{
  "generated_at": "2026-11-16T00:44:38.959Z",
  "seasons": [
    "2026-W29",
    "2026-W30",
    "2026-W31",
    "2026-W34",
    "2026-W35",
    "2026-W36",
    "2026-W37",
    "2026-W38",
    "2026-W39",
    "2026-W40",
    "2026-W41",
    "2026-W42",
    "2026-W43",
    "2026-W44",
    "2026-W45",
    "2026-W46",
    "2026-W47"
  ],
  "note": "Proved columns are signed by the grams that measured them, across every season. Prices here are each machine's signed size (CPUs, memory, card) at the network's metered price list (USD per million tokens = µUSD per second / tokens per second). Estimates are labelled and never signed. API prices and licences are cited from OpenRouter and the Hugging Face hub with the time they were read.",
  "usable_tokens_per_sec": 2,
  "counts": {
    "rows": 628,
    "proved": 347,
    "with_api": 341,
    "network_only": 111,
    "cheaper_than_api_somewhere": 6,
    "licence_known": 559
  },
  "model_counts": {
    "models": 279,
    "measured": 152,
    "sold_by_an_api": 152,
    "measured_and_no_api_sells_it": 124,
    "cheaper_here_than_the_api": 0,
    "too_big_for_every_machine": 45
  },
  "providers": [
    {
      "id": "a5be46",
      "did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
      "name": "Gram ·a5be46",
      "claimed": false,
      "online": false,
      "last_seen": "2026-11-05T06:26:43.387Z",
      "version": "9c25d0d",
      "region": {
        "code": "US_97058",
        "city": "The Dalles",
        "state": "Oregon"
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/nvidia/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2,
        "cpus": 4,
        "ram_gb": 31.346599578857422,
        "memory_bandwidth_gb_s": 0.7,
        "gpu": "Tesla T4",
        "vram_gb": 14.6,
        "gpu_tflops_fp16": 24.5,
        "disk_read_mb_s": 1931.4,
        "gpu_count": 2,
        "gpu_org": "nvidia",
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
          "clock_ghz": 2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 1024,
          "l3_mb": 38.5,
          "single_core_hashes_per_sec": 182.6,
          "multi_core_hashes_per_sec": 387.6,
          "sustained_ratio": 1.1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-05T03:28:40.669Z",
          "measured_in": "US_97058",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 0.7,
          "latency_ns": 114.5,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-05T03:28:42.267Z",
          "measured_in": "US_97058",
          "signed": true
        },
        "gpu": {
          "name": "Tesla T4",
          "memory_gb": 14.6,
          "fp16_tflops": 24.5,
          "watts": 25.8,
          "tflops_per_watt": 0.947,
          "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
          "measured_at": "2026-11-05T03:49:39.140Z",
          "measured_in": "US_97058",
          "signed": true
        },
        "storage": {
          "read_mb_s": 1931.4,
          "write_mb_s": 1.2,
          "random_read_iops": 120098,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-11-05T03:29:34.741Z",
          "measured_in": "US_97058",
          "signed": true
        }
      },
      "signed_runs": 94,
      "signed_runs_30d": 30,
      "builds_measured": 94,
      "best_results": 29,
      "first_signed_at": "2026-10-23T22:48:46.587Z",
      "last_signed_at": "2026-11-05T06:18:49.589Z",
      "awake_pct": 100,
      "queue": {
        "measured": 6,
        "failed": 18,
        "queued": 1521
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-29",
          "hours": 11
        },
        {
          "date": "2026-10-30",
          "hours": 3
        },
        {
          "date": "2026-11-02",
          "hours": 5
        },
        {
          "date": "2026-11-03",
          "hours": 9
        },
        {
          "date": "2026-11-04",
          "hours": 15
        },
        {
          "date": "2026-11-05",
          "hours": 7
        }
      ],
      "uptime_pct": 34.7,
      "best_value": {
        "model": "Qwen2.5 0.5B Instruct",
        "page": "/inferences/qwen-2-5/0.5b",
        "build": "Q4_K_M (4-bit) by Qwen",
        "build_page": "/inferences/qwen-2-5/0.5b/qwen-q4_k_m-ollama-148138f0",
        "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 239.08,
        "cost_usd_per_million": 0.5645,
        "measured_at": "2026-11-05T06:03:49.588Z",
        "context_tokens": 32768,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "pxVUn4UHtls7r7NfLMqDWbFXQTwsgW6WdNzz8mQKHskLrOx7SQqyjnO4j7CUcB4hSAeqJTh+WKugSIQ9lSkBBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "digHmAeSiq2yn9kBZ2Tid4dg+mWV1kevdfYhcsGVr8w="
        },
        "tokens_per_usd_hour": 492
      },
      "opened": [
        {
          "ref": "hf.co/microsoft/phi-4-gguf:BF16",
          "family": "phi-4",
          "member": "base",
          "at": "2026-10-23T22:48:46.587Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_L",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-05T06:18:49.589Z"
        },
        {
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q4_K_M",
          "family": "qwen-3",
          "member": "4b",
          "at": "2026-11-05T06:18:49.589Z"
        },
        {
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "family": "llama-3-2",
          "member": "3b",
          "at": "2026-11-05T06:18:49.589Z"
        },
        {
          "ref": "hf.co/QuantFactory/OpenHathi-7B-Hi-v0.1-Base-GGUF:Q4_K_M",
          "family": "openhathi",
          "member": "7b",
          "at": "2026-11-05T06:18:49.589Z"
        },
        {
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_K_M",
          "family": "qwen-3",
          "member": "4b",
          "at": "2026-11-05T06:18:49.589Z"
        },
        {
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q8_0",
          "family": "qwen-3",
          "member": "4b",
          "at": "2026-11-05T06:18:49.589Z"
        },
        {
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q8_K_XL",
          "family": "gpt-oss",
          "member": "20b",
          "at": "2026-11-05T06:18:49.589Z"
        },
        {
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ1_S",
          "family": "llama-3-2",
          "member": "3b",
          "at": "2026-11-05T06:18:49.589Z"
        },
        {
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:IQ3_M",
          "family": "llama-3-2",
          "member": "3b",
          "at": "2026-11-05T06:18:49.589Z"
        },
        {
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q3_K_L",
          "family": "llama-3-2",
          "member": "3b",
          "at": "2026-11-05T06:18:49.589Z"
        },
        {
          "ref": "hf.co/Qwen/Qwen3-VL-4B-Instruct-GGUF:F16",
          "family": "qwen-3",
          "member": "vl-4b",
          "at": "2026-11-05T06:18:49.589Z"
        },
        {
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q2_K",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-05T06:18:49.589Z"
        },
        {
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q3_K_M",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-05T06:18:49.589Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-23",
          "runs": 1
        },
        {
          "date": "2026-10-24",
          "runs": 39
        },
        {
          "date": "2026-10-29",
          "runs": 642
        },
        {
          "date": "2026-11-02",
          "runs": 303
        },
        {
          "date": "2026-11-03",
          "runs": 814
        },
        {
          "date": "2026-11-04",
          "runs": 1393
        },
        {
          "date": "2026-11-05",
          "runs": 669
        }
      ],
      "runs_by_family": [
        {
          "family": "gemma-4",
          "runs": 31
        },
        {
          "family": "qwen-3",
          "runs": 14
        },
        {
          "family": "phi-4",
          "runs": 11
        },
        {
          "family": "llama-3-2",
          "runs": 6
        },
        {
          "family": "sarvam",
          "runs": 3
        },
        {
          "family": "cogvideox",
          "runs": 3
        },
        {
          "family": "wan-2",
          "runs": 3
        },
        {
          "family": "musicgen",
          "runs": 2
        },
        {
          "family": "depth-anything",
          "runs": 2
        },
        {
          "family": "moonshine",
          "runs": 2
        },
        {
          "family": "gpt-oss",
          "runs": 2
        },
        {
          "family": "glm-4-7",
          "runs": 1
        },
        {
          "family": "sam",
          "runs": 1
        },
        {
          "family": "qwen-2-5",
          "runs": 1
        },
        {
          "family": "tiny-sd",
          "runs": 1
        },
        {
          "family": "sd-turbo",
          "runs": 1
        },
        {
          "family": "kokoro",
          "runs": 1
        },
        {
          "family": "whisper",
          "runs": 1
        },
        {
          "family": "sana",
          "runs": 1
        },
        {
          "family": "krutrim",
          "runs": 1
        },
        {
          "family": "airavata",
          "runs": 1
        },
        {
          "family": "openhathi",
          "runs": 1
        },
        {
          "family": "qwen-3-8",
          "runs": 1
        },
        {
          "family": "hunyuan-video",
          "runs": 1
        },
        {
          "family": "ltx-video",
          "runs": 1
        },
        {
          "family": "qwen-image",
          "runs": 1
        }
      ],
      "rooms": [
        "NL_9711",
        "US_20511",
        "US_51503",
        "US_97058"
      ],
      "awards": [
        {
          "id": "runs-10",
          "label": "10+ signed runs"
        },
        {
          "id": "opener",
          "label": "First to measure 14 builds"
        },
        {
          "id": "best-10",
          "label": "Best result on 29 builds"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "Qwen2.5 0.5B Instruct",
        "page": "/inferences/qwen-2-5/0.5b",
        "build": "Q4_K_M (4-bit) by Qwen",
        "build_page": "/inferences/qwen-2-5/0.5b/qwen-q4_k_m-ollama-148138f0",
        "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 239.08,
        "cost_usd_per_million": 0.5645,
        "measured_at": "2026-11-05T06:03:49.588Z",
        "context_tokens": 32768,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "pxVUn4UHtls7r7NfLMqDWbFXQTwsgW6WdNzz8mQKHskLrOx7SQqyjnO4j7CUcB4hSAeqJTh+WKugSIQ9lSkBBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "digHmAeSiq2yn9kBZ2Tid4dg+mWV1kevdfYhcsGVr8w="
        },
        "tokens_per_usd_hour": 492
      },
      "cheapest": {
        "model": "Qwen2.5 0.5B Instruct",
        "page": "/inferences/qwen-2-5/0.5b",
        "build": "Q4_K_M (4-bit) by Qwen",
        "build_page": "/inferences/qwen-2-5/0.5b/qwen-q4_k_m-ollama-148138f0",
        "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 239.08,
        "cost_usd_per_million": 0.5645,
        "measured_at": "2026-11-05T06:03:49.588Z",
        "context_tokens": 32768,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "pxVUn4UHtls7r7NfLMqDWbFXQTwsgW6WdNzz8mQKHskLrOx7SQqyjnO4j7CUcB4hSAeqJTh+WKugSIQ9lSkBBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "digHmAeSiq2yn9kBZ2Tid4dg+mWV1kevdfYhcsGVr8w="
        },
        "tokens_per_usd_hour": 492
      },
      "builds": [
        {
          "model": "Qwen2.5 0.5B Instruct",
          "page": "/inferences/qwen-2-5/0.5b",
          "build": "Q4_K_M (4-bit) by Qwen",
          "build_page": "/inferences/qwen-2-5/0.5b/qwen-q4_k_m-ollama-148138f0",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 239.08,
          "cost_usd_per_million": 0.5645,
          "measured_at": "2026-11-05T06:03:49.588Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "pxVUn4UHtls7r7NfLMqDWbFXQTwsgW6WdNzz8mQKHskLrOx7SQqyjnO4j7CUcB4hSAeqJTh+WKugSIQ9lSkBBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "digHmAeSiq2yn9kBZ2Tid4dg+mWV1kevdfYhcsGVr8w="
          },
          "tokens_per_usd_hour": 492
        },
        {
          "model": "Sarvam-1",
          "page": "/inferences/sarvam/1",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 111.22,
          "cost_usd_per_million": 1.2208,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "N8h+/CDdHI+jzwc8eWutheTSY09LHF7CXfHHdhkLHLWL7/oUNzEAKT2OcdnCa978pwaik/R1kFY0R26nTNBzDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "iFRMIC2eIdiwVAww3vrDwSrTUAaSSpE2Nj0vL7BXhpQ="
          },
          "tokens_per_usd_hour": 228
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "IQ1_S (1-bit) by unsloth",
          "build_page": "/inferences/llama-3-2/3b/unsloth-iq1_s-ollama",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ1_S",
          "engine": "ollama-host",
          "tokens_per_sec": 94.46,
          "cost_usd_per_million": 1.4323,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "1r8Qo78XIyntSJ4wux1aO10CbPqXylxbsVuuHVOb0MFYTyWCIkHpklSihye8QwdRgS/m8462q/NOHVPTZgZUCg==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6OTQuNDU4MDU3MTMzNzUxMywidW5pdCI6InRva2Vucy9zIiwidHMiOjE3OTM4NTk1Mjk1ODg4MDkzNzksIm5vZGVfZGlkIjoiZGlkOmVwbjowMDI0MDgwMTEyMjA0ZGI5Yjc2ZTcyOGNlMDVhMzVkNmVjN2Y1YjVmNmRmYTZhOTM3MzJmYWQ0MzZiNDQzNjA2ZWM1YzMwYTViZTQ2IiwiZXh0cmEiOnsiaGFyZHdhcmVfY2xhc3MiOiJsaW51eC9hbWQ2NC9udmlkaWEvMzJnIiwibW9kZWwiOiJoZi5jby91bnNsb3RoL0xsYW1hLTMuMi0zQi1JbnN0cnVjdC1HR1VGOklRMV9TIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjQ5LCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjEwNi40ODExMjMwMDAwMDAwMSwidG90YWxfdG9rZW5zIjoxMDA1OH19",
            "payload_sha256": "j9OBNkhFbWUVRyVCyHbsiHTg5IskTEA5krZMxk68qv4="
          },
          "tokens_per_usd_hour": 194
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by google",
          "build_page": "/inferences/gemma-4/e2b/google-q4_k_m-ollama",
          "ref": "gemma4:e2b",
          "engine": "ollama-host",
          "tokens_per_sec": 91.08,
          "cost_usd_per_million": 1.5416,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "531znaVOfUJSwuU7xv+fghDMIf2XaH4Mp9+ec4R259syh5BX4nZaFlgJYpFmDWgG7uuKivZI2+nI6VCQkrSgAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "QDi7SmOTdtP4NFpzBT3MLT4gBdmcMKjhNytNz6BEYIU="
          },
          "tokens_per_usd_hour": 180
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/llama-3-2/3b/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 86.87,
          "cost_usd_per_million": 1.5492,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "2deLKLJAE4JjduhZthR6zTfgy15tcK+Gt7LAYCIIUJ7Lzet3JHLTG4sdzOu0sSMwDC2ZWRrUzxcEOdSqKrObCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "DTtLVgvXJp5zJ9a/UxKKA3y1le+derUkKmQ0zq8nicY="
          },
          "tokens_per_usd_hour": 179
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_0 (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_0-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_0",
          "engine": "ollama-host",
          "tokens_per_sec": 83.82,
          "cost_usd_per_million": 1.6269,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "AFpY9+ffmcniz6Nfj3L9Jv+dM5zaMV9OC+iSbg3xPqxhsmkcpXRBfSGoUzH7920KsQ39B1O2ufNqqWKvcG2QAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "Lv5VaZ6/OgBL4iAb68CoDGUhBhUYlQSN3A7eEKpmVlU="
          },
          "tokens_per_usd_hour": 171
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "Q3_K_L (3-bit) by bartowski",
          "build_page": "/inferences/llama-3-2/3b/bartowski-q3_k_l-ollama",
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "tokens_per_sec": 78.82,
          "cost_usd_per_million": 1.7074,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "Wt6aa3sVC0haNDXyaZhxJWhYywD+k2aUhljsN52ngYwj27k5jMUzGDhizfPwzzBtCQw7cNIGGi1ZTxolT5lcCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "6P8cmNtxgglLJ0+2GlldMVkOIVOAm3S/r676PbCx2fM="
          },
          "tokens_per_usd_hour": 163
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "IQ4_XS (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-iq4_xs-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_XS",
          "engine": "ollama-host",
          "tokens_per_sec": 78.69,
          "cost_usd_per_million": 1.7397,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "UXXrqrsUXrQW86qz+sZdY2gIaJUOd2bt+qdnN0tfvxeA+LrYrQDmM43UIx2V8HaFWnzPKflXmz3RixcOKp1rDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "+SJCD5FFktWC4iZk9/sbteddYsESplj6RpN9Pwdl5IQ="
          },
          "tokens_per_usd_hour": 160
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "tokens_per_sec": 77.58,
          "cost_usd_per_million": 1.7757,
          "measured_at": "2026-10-19T11:28:03.450Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:00240801122095368b040a04cc8b46563ce9ac25e091edccd520d7753bb0a0093f9490e78fd2",
            "signature": "OVe3nttZDdDnZkX344L+tBDQQsFslXcX4T6fYZDm8AXTtAE2Ki6wp2w1ikvYhW/kC46HbqrTXOjoz9dpmpVQCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "Yn8g7FG6ZJeVRPyQI1hfyHm9MwQeGWZQjYqWcxW1z8M="
          },
          "tokens_per_usd_hour": 156
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 74.2,
          "cost_usd_per_million": 1.8399,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "kAbAwjkvdsBFOCbF1W8lfWET8F5ase9u7cUpkZyX+VlsI2O2767fj/bZL3UZ19P9g7wYf/hb8WyhITV69XjgBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "SEJrC5gK/PJphBAsMw/UpXXofHpoFPVGcKKiD6Il5mc="
          },
          "tokens_per_usd_hour": 151
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "IQ3_M (3-bit) by bartowski",
          "build_page": "/inferences/llama-3-2/3b/bartowski-iq3_m-ollama",
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:IQ3_M",
          "engine": "ollama-host",
          "tokens_per_sec": 71.97,
          "cost_usd_per_million": 1.87,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "y0C6GGftevjyABlHXofX+avXNsuSI7vPoLJHdsLPYF9sPXjQop28u5A4BIDCHNey6ljOZ0IroRIP54bAgZwDAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "89nLsZVL7Wh9Nppa07uqoVObsKhGvYx/hA/YnmkMiT4="
          },
          "tokens_per_usd_hour": 149
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q2_K (2-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini/unsloth-q2_k-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q2_K",
          "engine": "ollama-host",
          "tokens_per_sec": 68.61,
          "cost_usd_per_million": 1.9806,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "miWlotwURDVdfj8A9vdzRloamTA7+fI14TpV23WXNkOwY0VVhzEmoVcdW8q/IcwwQTwhCD1NggetH9V/Nyl+Cw==",
            "signing_payload_b64": "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",
            "payload_sha256": "4STfVZIT9Rn2TEV8q8YwHKcoYfwObI5ixfYMYGM/nps="
          },
          "tokens_per_usd_hour": 140
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q3_K_M (3-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini/unsloth-q3_k_m-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q3_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 66.57,
          "cost_usd_per_million": 2.0463,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "AKcFh//Njg8iVOysLcmxh9/yN2azllYq1iVbZqyTixUxcLAnN3gD4eXVa+FIK/xMlJV55iaE/L1aPYkXgq8XAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "yY3qtZpEj7hF/3GYz7gsTXh+0QIMVSH4/+S67ncU0a4="
          },
          "tokens_per_usd_hour": 136
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q3_K_L (3-bit) by maziyarpanahi",
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q3_k_l-ollama",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "tokens_per_sec": 64.06,
          "cost_usd_per_million": 2.1009,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "ddtE+Hqpxa/EFv19wHPL6wEkyjCIdjPLgGqlhFJrEBONKtMLdOT1tf01fDEcnkrONVRcLk+OFdzsOiduSPziAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "6VIQ5W2keskTi1PiCyFG5CGsOlYlFS4FVIot+eTbBT8="
          },
          "tokens_per_usd_hour": 132
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "IQ1_S (1-bit) by unsloth",
          "build_page": "/inferences/qwen-3/4b/unsloth-iq1_s-ollama",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ1_S",
          "engine": "ollama-host",
          "tokens_per_sec": 61.33,
          "cost_usd_per_million": 2.2082,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "LJXlh2taxmczw36A2OPKGSanjAHyzVHiKPpBJ+7cmTvkt0wfgJQ+R7QBExjK+qpfZT49mI6dORIf/8SkxNwlAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "q4DvIIdDBFVeIosDU+DaZPIRWL2/khEBEJqVuuY4Ym8="
          },
          "tokens_per_usd_hour": 126
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q3_K_S (3-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q3_k_s-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 54.73,
          "cost_usd_per_million": 2.6214,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 65536,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "6A+nYRoD6Sl7qIEUtL1lnWssfU8Lz08wJFCbGlNgM9wOLYTlTe1yQsF0KEsHk3CGmpdKxV9kkUTQWr/4uhJKBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "VVj3Xe2drzC4Fx6sjqk1xAQTL6t2FAWwoyOz2QTxMok="
          },
          "tokens_per_usd_hour": 106
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "Q4_K_M (4-bit) by Qwen",
          "build_page": "/inferences/qwen-3/4b/qwen-q4_k_m-ollama",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 50.44,
          "cost_usd_per_million": 2.7067,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 16384,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "nGqb3S0xTPGZ2AlIS2PLvtGBNbRGXG6AtA8OC+j9iOcITry7kctHeURIc1RMJ/zVIgWUphw6qxn2WY1FUMgXBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "ugBUxSE+6Hfp9Uwol9/XXPSQZdbcNuF4qSUzhRbBihI="
          },
          "tokens_per_usd_hour": 103
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "Q5_0 (5-bit) by Qwen",
          "build_page": "/inferences/qwen-3/4b/qwen-q5_0-ollama",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_0",
          "engine": "ollama-host",
          "tokens_per_sec": 47.96,
          "cost_usd_per_million": 2.8516,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "IiTdVcTiZlLFKD9QzCavw53zoVjVXtP3Ns3ULmIviVOkF4YbuoHQxZ4DXMumJD88JQaaVxAGv5LRfNSBt20pAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "DpT4y+Od9MtoPjXdD6G66NWoZUGTydRKE0X0f/fhVB8="
          },
          "tokens_per_usd_hour": 97
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "IQ3_XXS (3-bit) by unsloth",
          "build_page": "/inferences/qwen-3/4b/unsloth-iq3_xxs-ollama",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "tokens_per_sec": 47.81,
          "cost_usd_per_million": 2.8422,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "/pYo1R9QvwpOjh/soU8j/lqUNBXopn/VFaae7rMbtKmxqGVglsfJuqG2P7c0sc1rpBvsufyAYVM/K0RYR08UAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "0s1r4cFJ5N6j8La8jh4/kXaBuD2TdzZ5doA7Yt5Caug="
          },
          "tokens_per_usd_hour": 98
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "Q5_K_M (5-bit) by Qwen",
          "build_page": "/inferences/qwen-3/4b/qwen-q5_k_m-ollama",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 46.99,
          "cost_usd_per_million": 2.9117,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "rKSwONDCTRIrdAeMayWtGpPHP1LWWP2YGoGQcxPs7yic19JR7lEwcP3HiMdPKi4cdRLfizWG4p6QLT5M5JmKAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "QzCVTZCh9KyTySO6zx/xHv7hSuzkBgeDnWb74PwWaT0="
          },
          "tokens_per_usd_hour": 95
        },
        {
          "model": "GLM-4.7 Flash",
          "page": "/inferences/glm-4-7/flash",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/glm-4-7/flash/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/GLM-4.7-Flash-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 44.98,
          "cost_usd_per_million": 5.6235,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "XUEyipid46+kLBdt3vsYkzJSj7lYh1XmPL/r4O2WnoJJpl0lOCh1PyxxpoU9dCw5dAHz1NzB/nNmVCgSc8TTDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "TmLvZzzo9pfzI8v17eOhsuWijolaIu8VdUXR11005UQ="
          },
          "tokens_per_usd_hour": 49
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q8_K_XL (8-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q8_k_xl-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q8_K_XL",
          "engine": "ollama-host",
          "tokens_per_sec": 44.16,
          "cost_usd_per_million": 3.2797,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "oiUJV6I1RqoOEmrMo8b90V6Ri8LJeIqyOWgEz3ICAqJ3NBv9enH/SP1I0ZL9I+fBqFA1eb45GReehF+bldvECA==",
            "signing_payload_b64": "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",
            "payload_sha256": "jlOYTMHrjhM6BwDbbBd0wJWc/UgODgcIwx4JwdNq2js="
          },
          "tokens_per_usd_hour": 85
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "Q8_0 (8-bit) by Qwen",
          "build_page": "/inferences/qwen-3/4b/qwen-q8_0-ollama",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 38.86,
          "cost_usd_per_million": 3.5489,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "0mEJwHdK0Zw44csZYSklcgcS3ukop68JRN/ywbFIOXqupfLnFSW9ybbdNb+RAYNjgTnO8AboymvMQNNKM7YQBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "MfBBnhoVHnJ7vuytKpxd+qVXBJXDiDBe/CA2bvHwZsA="
          },
          "tokens_per_usd_hour": 78
        },
        {
          "model": "Airavata 7B (Hindi)",
          "page": "/inferences/airavata/7b",
          "build": "Q4_K_M (4-bit) by mradermacher",
          "build_page": "/inferences/airavata/7b/mradermacher-q4_k_m-ollama",
          "ref": "hf.co/mradermacher/Airavata-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 31.59,
          "cost_usd_per_million": 4.3623,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "W53EpXULaAA38Bq6UyKAJfIDNnsnH9CaKD1H+cTQGR8/uJoj8c4p0NzrS2qXnJO288dCbXK4DEIp6IijjrL7DQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "FY1TSt1LPBJBlwdEvfp/pRjW+fV1nlE5hUljaYLmrRI="
          },
          "tokens_per_usd_hour": 64
        },
        {
          "model": "OpenHathi 7B Hindi base",
          "page": "/inferences/openhathi/7b",
          "build": "Q4_K_M (4-bit) by QuantFactory",
          "build_page": "/inferences/openhathi/7b/quantfactory-q4_k_m-ollama",
          "ref": "hf.co/QuantFactory/OpenHathi-7B-Hi-v0.1-Base-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 29.84,
          "cost_usd_per_million": 4.6183,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "4IlGtATg8l/Qb5dr67qSNDegWE0907eRvB88tqfHR6Wlye9DQxogdwZUhStBF3KzthGy+viAubOFm+MKV7ySBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "i+c8QRQgYxf87SX+zxUUe0MvxKgquPAGO0Wm6FAseSE="
          },
          "tokens_per_usd_hour": 60
        },
        {
          "model": "Sarvam-M",
          "page": "/inferences/sarvam/m",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/sarvam/m/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/sarvamai_sarvam-m-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 15.61,
          "cost_usd_per_million": 9.3329,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "IiaX6zXGosC9aNd7a4vC5mBOFHuQ4TtpW1L7qTu4LCS4OYJpvvFmgn+fM6O1zpm3+4/gokwF2cXFjaR3htSKCQ==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTUuNjEyMjQzMTc1Mzg2MTc0LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5Mzg1OTUyOTU4ODgwOTM3OSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDRkYjliNzZlNzI4Y2UwNWEzNWQ2ZWM3ZjViNWY2ZGZhNmE5MzczMmZhZDQzNmI0NDM2MDZlYzVjMzBhNWJlNDYiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L252aWRpYS8zMmciLCJtb2RlbCI6ImhmLmNvL2JhcnRvd3NraS9zYXJ2YW1haV9zYXJ2YW0tbS1HR1VGOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo3Nywic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjoxNjk0Ljk1MTgyMDk5ODk5OTcsInRvdGFsX3Rva2VucyI6MjY0NjJ9fQ==",
            "payload_sha256": "FdGHr57sQjHOTEt8WGZd23fANyfBF53d8ZZhNz8q0BY="
          },
          "tokens_per_usd_hour": 30
        },
        {
          "model": "Krutrim-2 Instruct",
          "page": "/inferences/krutrim/2",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/krutrim/2/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/krutrim-ai-labs_Krutrim-2-instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 13.56,
          "cost_usd_per_million": 10.3505,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 65536,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "Ck44BL42uOhTnx5nwRTcDGqfSVXDnkqa5Et+jgp7GjEgU7G36ACxA+9AYdZVDfrIM/T2TTGrHYD+Y1XM4sDeDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "h5TODmxeM82uG4raHdhz2qwPrt/C3iegbEyTiCAf550="
          },
          "tokens_per_usd_hour": 27
        },
        {
          "model": "Qwen 3 VL-4B",
          "page": "/inferences/qwen-3/vl-4b",
          "build": "F16 (16-bit, full precision) by Qwen",
          "build_page": "/inferences/qwen-3/vl-4b/qwen-f16-ollama",
          "ref": "hf.co/Qwen/Qwen3-VL-4B-Instruct-GGUF:F16",
          "engine": "ollama-host",
          "tokens_per_sec": 11.91,
          "cost_usd_per_million": 11.8296,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "ZWHEb6YO7EFSsqq3NikNO5847dBPxFOCaQUtvujdik5Tp/hN8XGEoJfU6jKuGdZbHN9f7hyAmMZMwFymBG4qBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "7XR8VyEin+ccoQdtU1Zg2r1GrGL10a6IVI50QOOOHlI="
          },
          "tokens_per_usd_hour": 23
        },
        {
          "model": "Qwen 3.8 27B",
          "page": "/inferences/qwen-3-8/27b",
          "build": "Q8_0 (8-bit) by unsloth",
          "build_page": "/inferences/qwen-3-8/27b/unsloth-q8_0-ollama",
          "ref": "hf.co/unsloth/Qwen3.8-27B-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 2.77,
          "cost_usd_per_million": 94.304,
          "measured_at": "2026-11-05T06:18:49.588Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "NAgKl9Ub2Xiw2WykYiqFsP3ChrRz44sKSYXWIqB9klPPsBM6ElJDlpgWo3YWfcFIrhgIc5RXaZkprfOgk8f5Dw==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6Mi43NzA3NTEyNDI4OTA5NzUsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzODU5NTI5NTg4ODA5Mzc5LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwNGRiOWI3NmU3MjhjZTA1YTM1ZDZlYzdmNWI1ZjZkZmE2YTkzNzMyZmFkNDM2YjQ0MzYwNmVjNWMzMGE1YmU0NiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbnZpZGlhLzMyZyIsIm1vZGVsIjoiaGYuY28vdW5zbG90aC9Rd2VuMy44LTI3Qi1HR1VGOlE4XzAiLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6MjksInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6MTU5Ny43NjE2MjIsInRvdGFsX3Rva2VucyI6NDQyN319",
            "payload_sha256": "tZcwj+6t8SZuD7d484G+Pqse1jm8aakRqzhp0RIqDTo="
          },
          "tokens_per_usd_hour": 3
        }
      ],
      "rank": 1,
      "of": 36,
      "percentiles": {
        "gpu_tflops": 40,
        "memory_bandwidth": 15,
        "cpu_all_cores": 25,
        "cpu_one_core": 39,
        "disk_read": 29,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "e61d3b",
      "did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
      "name": "Gram ·e61d3b",
      "claimed": false,
      "online": false,
      "last_seen": "2026-10-31T08:12:35.161Z",
      "version": "453201c",
      "region": {
        "code": "NL_9711",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2.2,
        "cpus": 4,
        "ram_gb": 31.346603393554688,
        "memory_bandwidth_gb_s": 3.4,
        "gpu": "Tesla T4",
        "vram_gb": 14.6,
        "gpu_tflops_fp16": 11.5,
        "disk_read_mb_s": 4271.7,
        "gpu_count": 2,
        "gpu_org": "nvidia",
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
          "clock_ghz": 2.2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 256,
          "l3_mb": 55,
          "single_core_hashes_per_sec": 181,
          "multi_core_hashes_per_sec": 436.8,
          "sustained_ratio": 1.1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-10-30T20:43:55.306Z",
          "measured_in": "NL_9711",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 3.4,
          "latency_ns": 49.1,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-10-30T20:43:56.203Z",
          "measured_in": "NL_9711",
          "signed": true
        },
        "gpu": {
          "name": "Tesla T4",
          "memory_gb": 14.6,
          "fp16_tflops": 11.5,
          "watts": 105,
          "tflops_per_watt": 0.109,
          "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
          "measured_at": "2026-10-28T10:39:20.231Z",
          "measured_in": "SG_190464",
          "signed": true
        },
        "storage": {
          "read_mb_s": 4271.7,
          "write_mb_s": 177.9,
          "random_read_iops": 391634,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-10-30T20:43:56.799Z",
          "measured_in": "NL_9711",
          "signed": true
        }
      },
      "signed_runs": 93,
      "signed_runs_30d": 61,
      "builds_measured": 93,
      "best_results": 47,
      "first_signed_at": "2026-10-28T14:49:00.245Z",
      "last_signed_at": "2026-10-31T00:18:37.299Z",
      "awake_pct": 100,
      "queue": {
        "measured": 62,
        "failed": 24,
        "queued": 1299
      },
      "download_mib_per_sec": 54.3,
      "uptime_days": [
        {
          "date": "2026-10-29",
          "hours": 5
        },
        {
          "date": "2026-10-30",
          "hours": 23
        },
        {
          "date": "2026-10-31",
          "hours": 9
        }
      ],
      "uptime_pct": 51.4,
      "best_value": {
        "model": "Forge-1-Mini",
        "page": null,
        "build": null,
        "build_page": "/inferences/north-1/forge-1-mini~north-ml1/north-ml1-q3_k_m-ollama",
        "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q3_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 628.82,
        "cost_usd_per_million": 0.0484,
        "measured_at": "2026-10-31T00:18:37.298Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "y/9+gykKIPCSJsYSEpvk9fIJQnUY75Um7tOW/5UIHGkPdZkAh2zu/zo427J6FaA4rDBIJaC/lY5GI3LcftmICQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "9j8M6DoaTc5FGJkESYl9qArmKrzf2bhtRyay8YPHL34="
        },
        "tokens_per_usd_hour": 5743
      },
      "opened": [
        {
          "ref": "hf.co/unsloth/Qwen3.5-9B-GGUF:Q4_K_M",
          "family": "qwen-3-5",
          "member": "9b",
          "at": "2026-10-30T04:00:04.405Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:Q8_0",
          "family": "gemma-4",
          "member": "26b-a4b",
          "at": "2026-10-30T05:17:09.503Z"
        },
        {
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
          "family": "llama-3-2",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:IQ1_S",
          "family": "llama-3-2",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/ibm-granite/granite-4.2-3b-GGUF:Q4_K_M",
          "family": "granite-4-2",
          "member": "3b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_XS",
          "family": "gemma-4",
          "member": "e2b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:F16",
          "family": "llama-3-2",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/mradermacher/HY-MT1.5-1.8B-i1-GGUF:Q4_0",
          "family": "hy-mt-1-5",
          "member": "1.8b~mradermacher-i1",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-4-12b-it-GGUF:Q4_K_M",
          "family": "gemma-4",
          "member": "12b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:BF16",
          "family": "gemma-4",
          "member": "e2b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q8_0",
          "family": "gemma-4",
          "member": "e2b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3.8-27B-GGUF:Q8_0",
          "family": "qwen-3-8",
          "member": "27b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:IQ3_M",
          "family": "llama-3-2",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q3_K_L",
          "family": "llama-3-2",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q8_0",
          "family": "llama-3-2",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/DavidAU/LFM2.5-2.6B-Qwen3.8-Turbo-Brilliance-Power-X12-NEO-MAX-GGUF:Q8_0",
          "family": "lfm-2-5",
          "member": "2.6b~davidau-qwen3-8-turbo-brilliance-power-x12-neo-max",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/ggml-org/gemma-3-1b-it-GGUF:Q8_0",
          "family": "gemma-3",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/HauhauCS/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP-GGUF:Q6_K_P",
          "family": "qwen-3-8",
          "member": "27b~hauhaucs-uncensored-aggressive-mtp",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/hugging-quants/Llama-3.2-1B-Instruct-Q4_K_M-GGUF:Q4_K_M",
          "family": "llama-3-2",
          "member": "1b~hugging-quants-llama-3-2-k-m",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF:Q4_K_M",
          "family": "qwen-3-8",
          "member": "27b~huihui-ai-abliterated",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF:Q4_K_S",
          "family": "qwen-3-8",
          "member": "27b~huihui-ai-abliterated",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:BF16",
          "family": "granite-4-0",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q2_K",
          "family": "granite-4-0",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q3_K_L",
          "family": "granite-4-0",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q3_K_S",
          "family": "granite-4-0",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q4_0",
          "family": "granite-4-0",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/ibm-granite/granite-4.2-30b-GGUF:Q4_K_M",
          "family": "granite-4-2",
          "member": "30b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/ibm-granite/granite-4.2-8b-GGUF:Q4_K_M",
          "family": "granite-4-2",
          "member": "8b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/JonathanColetti/Qwen3.8-27B-Uncensored-GGUF:Q4_K_M",
          "family": "qwen-3-8",
          "member": "27b~jonathancoletti-uncensored",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/JonathanColetti/Qwen3.8-27B-Uncensored-GGUF:Q5_K_M",
          "family": "qwen-3-8",
          "member": "27b~jonathancoletti-uncensored",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/lmstudio-community/gemma-3-1b-it-GGUF:Q3_K_L",
          "family": "gemma-3",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/lmstudio-community/gemma-3-1b-it-GGUF:Q6_K",
          "family": "gemma-3",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/gemma-3-1b-it-GGUF:Q2_K",
          "family": "gemma-3",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/gemma-3-1b-it-GGUF:Q3_K_S",
          "family": "gemma-3",
          "member": "1b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/mradermacher/Kimi-Linear-REAP-35B-A3B-Instruct-i1-GGUF:Q4_K_M",
          "family": "kimi-linear",
          "member": "35b-a3b~cerebras-reap",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/mradermacher/MiniMax-M2.1-REAP-50-i1-GGUF:IQ1_S",
          "family": "minimax-m2-1",
          "member": "reap-50-i1",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q3_K_M",
          "family": "north-1",
          "member": "forge-1-mini~north-ml1",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/peculiar-ragdoll/Dirk-Qwen3.8-27B-GGUF:Q4_K_M",
          "family": "qwen-3-8",
          "member": "27b~peculiar-ragdoll-dirk",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/Qwen/Qwen3-0.6B-GGUF:Q8_0",
          "family": "qwen-3",
          "member": "0.6b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/Qwen/Qwen3-1.7B-GGUF:Q8_0",
          "family": "qwen-3",
          "member": "1.7b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/tvall43/Qwen3.6-14B-A3B-FableVibes-GGUF:Q4_K_M",
          "family": "qwen-3-6",
          "member": "14b-a3b~tvall43-fablevibes",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/unsloth/LFM2.5-24B-A2B-GGUF:IQ3_XXS",
          "family": "lfm-2-5",
          "member": "24b-a2b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q4_0",
          "family": "qwen-3-5",
          "member": "0.8b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q8_0",
          "family": "qwen-3-5",
          "member": "0.8b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3.5-2B-GGUF:Q4_0",
          "family": "qwen-3-5",
          "member": "2b",
          "at": "2026-10-31T00:18:37.299Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3.5-2B-GGUF:Q8_0",
          "family": "qwen-3-5",
          "member": "2b",
          "at": "2026-10-31T00:18:37.299Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-28",
          "runs": 52
        },
        {
          "date": "2026-10-29",
          "runs": 700
        },
        {
          "date": "2026-10-30",
          "runs": 2382
        },
        {
          "date": "2026-10-31",
          "runs": 120
        }
      ],
      "runs_by_family": [
        {
          "family": "embeddinggemma",
          "runs": 13
        },
        {
          "family": "llama-3-2",
          "runs": 8
        },
        {
          "family": "gemma-4",
          "runs": 8
        },
        {
          "family": "qwen-3-5",
          "runs": 8
        },
        {
          "family": "qwen-3-8",
          "runs": 8
        },
        {
          "family": "gemma-3",
          "runs": 6
        },
        {
          "family": "granite-4-0",
          "runs": 6
        },
        {
          "family": "granite-4-2",
          "runs": 3
        },
        {
          "family": "depth-anything",
          "runs": 2
        },
        {
          "family": "sd-turbo",
          "runs": 2
        },
        {
          "family": "moonshine",
          "runs": 2
        },
        {
          "family": "qwen-3-6",
          "runs": 2
        },
        {
          "family": "kimi-linear",
          "runs": 2
        },
        {
          "family": "north-1",
          "runs": 2
        },
        {
          "family": "lfm-2-5",
          "runs": 2
        },
        {
          "family": "qwen-3",
          "runs": 2
        },
        {
          "family": "ernie-4-5",
          "runs": 2
        },
        {
          "family": "glm-4-7",
          "runs": 1
        },
        {
          "family": "sam",
          "runs": 1
        },
        {
          "family": "qwen-2-5",
          "runs": 1
        },
        {
          "family": "nllb",
          "runs": 1
        },
        {
          "family": "tiny-sd",
          "runs": 1
        },
        {
          "family": "gpt-oss",
          "runs": 1
        },
        {
          "family": "kokoro",
          "runs": 1
        },
        {
          "family": "whisper",
          "runs": 1
        },
        {
          "family": "f5-tts",
          "runs": 1
        },
        {
          "family": "muse-glimmer",
          "runs": 1
        },
        {
          "family": "hy-mt-1-5",
          "runs": 1
        },
        {
          "family": "hy-mt-2",
          "runs": 1
        },
        {
          "family": "ministral-3",
          "runs": 1
        },
        {
          "family": "minimax-m2-1",
          "runs": 1
        },
        {
          "family": "whisper-indic-ft",
          "runs": 1
        }
      ],
      "rooms": [
        "NL_9711",
        "SG_190464",
        "US_29415",
        "US_51503"
      ],
      "awards": [
        {
          "id": "runs-10",
          "label": "10+ signed runs"
        },
        {
          "id": "opener",
          "label": "First to measure 46 builds"
        },
        {
          "id": "best-10",
          "label": "Best result on 47 builds"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "Forge-1-Mini",
        "page": null,
        "build": null,
        "build_page": "/inferences/north-1/forge-1-mini~north-ml1/north-ml1-q3_k_m-ollama",
        "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q3_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 628.82,
        "cost_usd_per_million": 0.0484,
        "measured_at": "2026-10-31T00:18:37.298Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "y/9+gykKIPCSJsYSEpvk9fIJQnUY75Um7tOW/5UIHGkPdZkAh2zu/zo427J6FaA4rDBIJaC/lY5GI3LcftmICQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "9j8M6DoaTc5FGJkESYl9qArmKrzf2bhtRyay8YPHL34="
        },
        "tokens_per_usd_hour": 5743
      },
      "cheapest": {
        "model": "Forge-1-Mini",
        "page": null,
        "build": null,
        "build_page": "/inferences/north-1/forge-1-mini~north-ml1/north-ml1-q3_k_m-ollama",
        "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q3_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 628.82,
        "cost_usd_per_million": 0.0484,
        "measured_at": "2026-10-31T00:18:37.298Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "y/9+gykKIPCSJsYSEpvk9fIJQnUY75Um7tOW/5UIHGkPdZkAh2zu/zo427J6FaA4rDBIJaC/lY5GI3LcftmICQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "9j8M6DoaTc5FGJkESYl9qArmKrzf2bhtRyay8YPHL34="
        },
        "tokens_per_usd_hour": 5743
      },
      "builds": [
        {
          "model": "Forge-1-Mini",
          "page": null,
          "build": null,
          "build_page": "/inferences/north-1/forge-1-mini~north-ml1/north-ml1-q3_k_m-ollama",
          "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q3_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 628.82,
          "cost_usd_per_million": 0.0484,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "y/9+gykKIPCSJsYSEpvk9fIJQnUY75Um7tOW/5UIHGkPdZkAh2zu/zo427J6FaA4rDBIJaC/lY5GI3LcftmICQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "9j8M6DoaTc5FGJkESYl9qArmKrzf2bhtRyay8YPHL34="
          },
          "tokens_per_usd_hour": 5743
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "tokens_per_sec": 24.89,
          "cost_usd_per_million": 1.3495,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "xoMYQdk2S6coNrQoSe48y4BUZ1DYdZQPvF/hlFNFm69VxgaaL9EQswDkBgX8bbbmrI1J40aGYzNpQDY18pk8DA==",
            "signing_payload_b64": "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",
            "payload_sha256": "+XON2DT74QDrtvHe/FF4lw7Xl04ZAIijAFNy+D2QUhw="
          },
          "tokens_per_usd_hour": 206
        },
        {
          "model": "Qwen2.5 0.5B Instruct",
          "page": "/inferences/qwen-2-5/0.5b",
          "build": "Q4_K_M (4-bit) by Qwen",
          "build_page": "/inferences/qwen-2-5/0.5b/qwen-q4_k_m-ollama-148138f0",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 18.98,
          "cost_usd_per_million": 1.6229,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "D3oUPZtVj7YvoxG5YoLjPtEwJxK9MW0KQix7k9HPcOamJDUv1P/HriImF8RVo9Nfg01X//eWa1uFJk1UhRhiAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "fNGm9V5lYJl9qjzmctRr7iCus7qGJi5uOlSp26dxKyo="
          },
          "tokens_per_usd_hour": 171
        },
        {
          "model": "Qwen 3.5 0.8B",
          "page": "/inferences/qwen-3-5/0.8b",
          "build": "Q4_0 (4-bit) by unsloth",
          "build_page": "/inferences/qwen-3-5/0.8b/unsloth-q4_0-ollama",
          "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q4_0",
          "engine": "ollama-host",
          "tokens_per_sec": 16.53,
          "cost_usd_per_million": 1.8643,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "u2R+dnlRZMYS5P/+SLQJPwsSVs7HYAfMXIXJuvX1i3HO7LLuM7qeL+LQt3dareAvXcJQPlCnAoWRNmTAAjOLCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "2szx392CzPrvCGNMacZa2y5IAtUplu2xUq8IwpLzMVs="
          },
          "tokens_per_usd_hour": 149
        },
        {
          "model": "Qwen 3.5 0.8B",
          "page": "/inferences/qwen-3-5/0.8b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/qwen-3-5/0.8b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 16.1,
          "cost_usd_per_million": 1.9149,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "gpBFJYw0a12RTzr4EKF+YxLNRyADxiuwj3z07LFVxtLOedP8V7DQoKBj/ug55LirhvQkU+pFsWXefqxY3eaCCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "nnvBePZD6Tlgrf/l+r+C7g/kTd59hShv4VjF3yoC740="
          },
          "tokens_per_usd_hour": 145
        },
        {
          "model": "Gemma 3 1B",
          "page": "/inferences/gemma-3/1b",
          "build": "Q4_K_M (4-bit) by ggml-org",
          "build_page": "/inferences/gemma-3/1b/ggml-org-q4_k_m-ollama",
          "ref": "hf.co/ggml-org/gemma-3-1b-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 13.98,
          "cost_usd_per_million": 2.2199,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "yP8sNk1aVG8ndDXE3hGWT1Lp8lzjKWxYX6FjqZf2L7l06/ZqPmc5po2O5nITnjxVhoBkid68D+gbE8zS0zxDCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "SGmJiRvuTNS7M/OJOAxNYjDLOspJvqZkROzm3E8XXD8="
          },
          "tokens_per_usd_hour": 125
        },
        {
          "model": "Gemma 3 1B",
          "page": "/inferences/gemma-3/1b",
          "build": "Q3_K_S (3-bit) by maziyarpanahi",
          "build_page": "/inferences/gemma-3/1b/maziyarpanahi-q3_k_s-ollama",
          "ref": "hf.co/MaziyarPanahi/gemma-3-1b-it-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 13.61,
          "cost_usd_per_million": 2.2346,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "7yl4afdSdOMaQEsDrFBItwCgR4GCyYGr67MTFtbDE7buifN3RQkIckux4hO+LYjfWSNcKqK3cZ6H8j3YykImDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "mmew07fzANs2QicBDyy8uRTn+q/ZdJeIAZONj+ZMVVs="
          },
          "tokens_per_usd_hour": 124
        },
        {
          "model": "Gemma 3 1B",
          "page": "/inferences/gemma-3/1b",
          "build": "Q3_K_L (3-bit) by lmstudio-community",
          "build_page": "/inferences/gemma-3/1b/lmstudio-community-q3_k_l-ollama",
          "ref": "hf.co/lmstudio-community/gemma-3-1b-it-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "tokens_per_sec": 13.1,
          "cost_usd_per_million": 2.3671,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "7spjQg07k42DaGzFulWHl1FioLZLS45sDztgcXDq9tRDL4EDDbW1nG/Hmt1ESeW22ExSVhsuHhVsqiimgcXOCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "8oj2rt9Vxwh9EaLvJ8VagyX9vARLb8a0XYzFdubYuww="
          },
          "tokens_per_usd_hour": 117
        },
        {
          "model": "Gemma 3 1B",
          "page": "/inferences/gemma-3/1b",
          "build": "Q2_K (2-bit) by maziyarpanahi",
          "build_page": "/inferences/gemma-3/1b/maziyarpanahi-q2_k-ollama",
          "ref": "hf.co/MaziyarPanahi/gemma-3-1b-it-GGUF:Q2_K",
          "engine": "ollama-host",
          "tokens_per_sec": 12.96,
          "cost_usd_per_million": 2.3461,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "kw84W7bw8C7vH87eJ9SZjXef38Xu67Cz+eVr65sYi9pM7WSJcJc4zb7pshzl4/tdsRbaHSSsWEOJdoC+t01KBQ==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTIuOTY0Nzk4NTg5MzI4MDMyLCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5MzQwNTkxNzI5ODY0ODg5MSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMGVlMzVkYzQ2YTIyMTJjYjMxNDQ0Y2E0YzE5NjY5YWM4ZDk3YjA5MjUyMGI4YzllNGQyOWFiNDhkOWVlNjFkM2IiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L25vbmUvMzJnIiwibW9kZWwiOiJoZi5jby9NYXppeWFyUGFuYWhpL2dlbW1hLTMtMWItaXQtR0dVRjpRMl9LIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjQ5LCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjIzNS41NjA5Mjg5OTk5OTk5NiwidG90YWxfdG9rZW5zIjozMDU0fX0=",
            "payload_sha256": "XoFBbY0HlHb4doKHQf03LXkNUgTK3EhUmmd6pWZjYFw="
          },
          "tokens_per_usd_hour": 118
        },
        {
          "model": "Qwen 3 0.6B",
          "page": "/inferences/qwen-3/0.6b",
          "build": "Q8_0 (8-bit) by Qwen",
          "build_page": "/inferences/qwen-3/0.6b/qwen-q8_0-ollama",
          "ref": "hf.co/Qwen/Qwen3-0.6B-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 12.79,
          "cost_usd_per_million": 2.4165,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "niteGoX8PYMVwf9p1TUNZ+ozgvTdbaqpwKJZjOxEaXGx6nRadPEwy6sReyDKL0kLk4kOSo57EH7cWqQCIwS6DA==",
            "signing_payload_b64": "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",
            "payload_sha256": "yv+sghfDR8TIwnGaA2+/5F4A0PTXQFHb7SWc3NQdQ6k="
          },
          "tokens_per_usd_hour": 115
        },
        {
          "model": "Qwen 3.5 0.8B",
          "page": "/inferences/qwen-3-5/0.8b",
          "build": "Q8_0 (8-bit) by unsloth",
          "build_page": "/inferences/qwen-3-5/0.8b/unsloth-q8_0-ollama",
          "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 12.25,
          "cost_usd_per_million": 2.5338,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "AaLFtoydR0QYGv+ycnTK3IeSCIt6jZDhKWG1iRBzyVAug6ElXPPAs92XV0f0h07kacGNuDtNAWCcxsb+TXH0DA==",
            "signing_payload_b64": "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",
            "payload_sha256": "uKuG1FEqJOK0rq5bbvISFGdQmKyonqYmGPEFn2cdxCo="
          },
          "tokens_per_usd_hour": 110
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/llama-3-2/1b/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 12.24,
          "cost_usd_per_million": 2.5368,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "pd7F5+EgRJwYCbecjdb4Rie4hVhcSZJQR76WRWXwVJ832fw1PUpZTJ5WZ/aGo3bYLeKVbQ/1W30O+1LmBYqnBg==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTIuMjM3NTg4ODg4MTIzMjA4LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5MzQwNTkxNzI5ODY0ODg5MSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMGVlMzVkYzQ2YTIyMTJjYjMxNDQ0Y2E0YzE5NjY5YWM4ZDk3YjA5MjUyMGI4YzllNGQyOWFiNDhkOWVlNjFkM2IiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L25vbmUvMzJnIiwibW9kZWwiOiJoZi5jby9iYXJ0b3dza2kvTGxhbWEtMy4yLTFCLUluc3RydWN0LUdHVUY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjUyLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjIwOC43MDEyNDE5OTcsInRvdGFsX3Rva2VucyI6MjU1NH19",
            "payload_sha256": "fzkWkaVB3SPSyX/ndXuuGiweZ7ZMM7nskja753MYi9M="
          },
          "tokens_per_usd_hour": 109
        },
        {
          "model": "Huihui-Qwen3.8-27B-abliterated-GGUF",
          "page": "/inferences/qwen-3-8/27b~huihui-ai-abliterated",
          "build": "Q4_K_S (4-bit) by huihui-ai",
          "build_page": "/inferences/qwen-3-8/27b~huihui-ai-abliterated/huihui-ai-q4_k_s-ollama",
          "ref": "hf.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF:Q4_K_S",
          "engine": "ollama",
          "tokens_per_sec": 11.84,
          "cost_usd_per_million": null,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "X4KqT2b8Ux+rekeKsO/Qed9hq7p+oXl6si2jkg+Bn5OOiYqa5PHjgwPSF/SbsmP/mKfmxGW1SyE4ogawd+sEDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "GncILOtDomyzVF4GcI3EcLZRWDxzwYHObfI2tWnvYzM="
          },
          "tokens_per_usd_hour": null
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": null,
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "tokens_per_sec": 11.45,
          "cost_usd_per_million": 2.6566,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "VwsbNGAF8UjFxH5JNbn4FCMEzEO4kR9zn//+9cHV5sEi0nm5UvPBDwURrb4XElNl1cc5+12lUWkUOJsX0y2aCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "qrUwcerxoHklzLicmUtNKQ8zcVGaGQaMF19KWfgt82o="
          },
          "tokens_per_usd_hour": 105
        },
        {
          "model": "Granite 4.0 1B",
          "page": "/inferences/granite-4-0/1b",
          "build": "Q4_K_M (4-bit) by ibm-granite",
          "build_page": "/inferences/granite-4-0/1b/ibm-granite-q4_k_m-ollama",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 11.37,
          "cost_usd_per_million": 2.744,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "nqLXOqxDriaxROVdKEjJzh4w1AwdH4z9t9XISaALgYLFWL/lVJv/M/CrIMEy3TEDbtrarVVB0DKziBwXiteQDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "z76Hz4HoYUep4hYU8qw84UMdoj1i6aybsAqYhxCvglw="
          },
          "tokens_per_usd_hour": 101
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": "Q4_K_M (4-bit) by hugging-quants",
          "build_page": "/inferences/llama-3-2/1b~hugging-quants-llama-3-2-k-m/hugging-quants-q4_k_m-ollama",
          "ref": "hf.co/hugging-quants/Llama-3.2-1B-Instruct-Q4_K_M-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 11.25,
          "cost_usd_per_million": 2.7606,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "fwWkqvBRrU4a78QT3IF6wnRaEuR1/94JWriCWzNq90enQ+WTd4zXOWK4Y5fgAlQ/Ua0iX1fYtUsEI/ieMlKNAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "gmkdtergC55JVGdKpWdG0gNXdvS8Yl6YsEV/qvCCYaU="
          },
          "tokens_per_usd_hour": 101
        },
        {
          "model": "Gemma 3 1B",
          "page": "/inferences/gemma-3/1b",
          "build": "Q6_K (6-bit) by lmstudio-community",
          "build_page": "/inferences/gemma-3/1b/lmstudio-community-q6_k-ollama",
          "ref": "hf.co/lmstudio-community/gemma-3-1b-it-GGUF:Q6_K",
          "engine": "ollama-host",
          "tokens_per_sec": 10.79,
          "cost_usd_per_million": 2.8924,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "s91a8yr+9Xwwje9YtTK4zQs7c/X0+3YUQvCjx+FEPlZOmDsmQucLtsx5yQIPGBDAve9itNYY0ram5h/NN75+AA==",
            "signing_payload_b64": "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",
            "payload_sha256": "WvQXoojc7FNivdkVG+IK6CsiGLwRfoIR14+nJYgSwrk="
          },
          "tokens_per_usd_hour": 96
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": "Q3_K_L (3-bit) by bartowski",
          "build_page": "/inferences/llama-3-2/1b/bartowski-q3_k_l-ollama",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "tokens_per_sec": 10.3,
          "cost_usd_per_million": 3.007,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "y3BjTARSXdwp5GFW5MZgPMg80ndl98HpL8EynpK2VZh5x6CWI0x6Vu/2kydrSuJ0mpE/PvJPg3ajE2jMy0LvDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "gujLB37UrpZPYg5xdCMPUao7DFHZJGZEgAZyHvZdHpw="
          },
          "tokens_per_usd_hour": 92
        },
        {
          "model": "Dirk-Qwen3.8-27B-GGUF",
          "page": "/inferences/qwen-3-8/27b~peculiar-ragdoll-dirk",
          "build": "Q4_K_M (4-bit) by peculiar-ragdoll",
          "build_page": "/inferences/qwen-3-8/27b~peculiar-ragdoll-dirk/peculiar-ragdoll-q4_k_m-ollama",
          "ref": "hf.co/peculiar-ragdoll/Dirk-Qwen3.8-27B-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 9.96,
          "cost_usd_per_million": null,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "pQ4N6+DceuMzsSzHoKxRJ6/fEiCVCr4YZRPe1zSudDD1ULwYrANE2dFD8lCkFNPUJxenj18+icYYFbrgs0V2CA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6OS45NTk5MTk1NDkwNjA2MzYsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzNDA1OTE3Mjk4NjQ4ODkxLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwZWUzNWRjNDZhMjIxMmNiMzE0NDRjYTRjMTk2NjlhYzhkOTdiMDkyNTIwYjhjOWU0ZDI5YWI0OGQ5ZWU2MWQzYiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbm9uZS8zMmciLCJtb2RlbCI6ImhmLmNvL3BlY3VsaWFyLXJhZ2RvbGwvRGlyay1Rd2VuMy44LTI3Qi1HR1VGOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo0Mywic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjozNDUuODg2MzI4MDAwMDAwMDUsInRvdGFsX3Rva2VucyI6MzQ0NX19",
            "payload_sha256": "VffiTnQel940rVCn8SyIKzcmTb/d0Ski2syMkcnRrjY="
          },
          "tokens_per_usd_hour": null
        },
        {
          "model": "Qwen3.8-27B-Uncensored-GGUF",
          "page": "/inferences/qwen-3-8/27b~jonathancoletti-uncensored",
          "build": "Q4_K_M (4-bit) by JonathanColetti",
          "build_page": "/inferences/qwen-3-8/27b~jonathancoletti-uncensored/jonathancoletti-q4_k_m-ollama",
          "ref": "hf.co/JonathanColetti/Qwen3.8-27B-Uncensored-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 9.85,
          "cost_usd_per_million": null,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "l1UIN1vv+4GJrhT006hy9Z5HYchMPUEYPmEIJ+wM77WD9lSUcZym2OSLsM1WR6B4wjM4dul3SkSnDLLUbOUpAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "mspraICjmgmzpLjarzj2YZZUFQAd97XDm10dEFTVb+4="
          },
          "tokens_per_usd_hour": null
        },
        {
          "model": "Gemma 3 1B",
          "page": "/inferences/gemma-3/1b",
          "build": "Q8_0 (8-bit) by ggml-org",
          "build_page": "/inferences/gemma-3/1b/ggml-org-q8_0-ollama",
          "ref": "hf.co/ggml-org/gemma-3-1b-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 9.53,
          "cost_usd_per_million": 3.2791,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "VThqWMXZ43YMZ9HbSf/Eg9DqFmQapYGT8W2ZiunK7qyWu01meHtxMao9XGVIswa8F8rX7cYuzaV5NOZDrnMyCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "jNol40QIbdH55GXg0kkQWmcnsv5S4lW1KdlXb/ZjnyE="
          },
          "tokens_per_usd_hour": 85
        },
        {
          "model": "Huihui-Qwen3.8-27B-abliterated-GGUF",
          "page": "/inferences/qwen-3-8/27b~huihui-ai-abliterated",
          "build": "Q4_K_M (4-bit) by huihui-ai",
          "build_page": "/inferences/qwen-3-8/27b~huihui-ai-abliterated/huihui-ai-q4_k_m-ollama",
          "ref": "hf.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 9.35,
          "cost_usd_per_million": null,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "IdYd1QDlrgDR6jdw2xzbkM/916tQSNPJNokzK4kRdANU6f+ZwfFa3Zw7d5/uIQG1RRMQ9jnMARCh5rWrmL/JDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "J9OZyIhKBej42Qn/uik06CN2eFL9tIgTf4IPGsMlDbs="
          },
          "tokens_per_usd_hour": null
        },
        {
          "model": "Granite 4.0 1B",
          "page": "/inferences/granite-4-0/1b",
          "build": "Q4_0 (4-bit) by ibm-granite",
          "build_page": "/inferences/granite-4-0/1b/ibm-granite-q4_0-ollama",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q4_0",
          "engine": "ollama-host",
          "tokens_per_sec": 8.75,
          "cost_usd_per_million": 3.5635,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "GEOfxTYMpxeor6q4+IwRuKk/oPADCFK8sbl3+h3qNUu3ESeGs1KWE6HdgleZZr49dlBFsyc5bqC5VH3Ed6PsBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "5pfYcU6+ABp+AEDd0l4HeX6Lhhj9cVi40eshalQ7fc4="
          },
          "tokens_per_usd_hour": 78
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": "Q8_0 (8-bit) by bartowski",
          "build_page": "/inferences/llama-3-2/1b/bartowski-q8_0-ollama",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 8.46,
          "cost_usd_per_million": 3.717,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "EOD1LKNsSNZ16rOZ1H2o0TdhPtv6al2Pd8YwBzwROpz+9QI7eLVZst9csdiNzJCwk8qe3UgCYPbaxXHAQnBuCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "M0zi1Gt9fv2vG0Lq37aDGZxMdA5KHkN4yTta1kv2Rs0="
          },
          "tokens_per_usd_hour": 75
        },
        {
          "model": "Granite 4.0 1B",
          "page": "/inferences/granite-4-0/1b",
          "build": "Q3_K_S (3-bit) by ibm-granite",
          "build_page": "/inferences/granite-4-0/1b/ibm-granite-q3_k_s-ollama",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 7.98,
          "cost_usd_per_million": 3.8867,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "gxcNkCZgHM72QnBInh+ro9oljDIeN9A+P2BwWMmbEHZcOwWxMbT/tkjZ1qR3zGnuanGOdXYhR8+n/bLShrA+Bw==",
            "signing_payload_b64": "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",
            "payload_sha256": "eo6WyNhxO2RPRQZj5nnbQpqREXZh+HQ4ucVSgkICrlk="
          },
          "tokens_per_usd_hour": 71
        },
        {
          "model": "Granite 4.0 1B",
          "page": "/inferences/granite-4-0/1b",
          "build": "Q2_K (2-bit) by ibm-granite",
          "build_page": "/inferences/granite-4-0/1b/ibm-granite-q2_k-ollama",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q2_K",
          "engine": "ollama-host",
          "tokens_per_sec": 7.82,
          "cost_usd_per_million": 3.9609,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "pI2VwoF8V8R/ThMzzxlzdzd2ORlDfk9hMwb3Sj3B8VmHUbyhuwf1Ar9jxmYGVEtZXHxD+vM3PBtdrTOuY3DXCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "qXp3RDYKZBeCgKtHUmVkt9wJbxxTPCENWgTFNWmXqVc="
          },
          "tokens_per_usd_hour": 70
        },
        {
          "model": "Granite 4.0 1B",
          "page": "/inferences/granite-4-0/1b",
          "build": "Q3_K_L (3-bit) by ibm-granite",
          "build_page": "/inferences/granite-4-0/1b/ibm-granite-q3_k_l-ollama",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "tokens_per_sec": 7.23,
          "cost_usd_per_million": 4.3044,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "tqOKISwjQs0EocUz7i5Z0jPjTAwZHUxgsxZkuVU3fsOH0uvqyE3aD9LbcDAvCfr6scIn/pPRO32ALEQZrRL7DA==",
            "signing_payload_b64": "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",
            "payload_sha256": "8DTR04zMbtp0f+Op6R5gLMeyTnnG80GEGvOKyeeR5P0="
          },
          "tokens_per_usd_hour": 64
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by google",
          "build_page": "/inferences/gemma-4/e2b/google-q4_k_m-ollama",
          "ref": "gemma4:e2b",
          "engine": "ollama-host",
          "tokens_per_sec": 7.15,
          "cost_usd_per_million": 5.0695,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "XryWV44xDp4vYZ6l4mKLHXmvcI147mlFr5Ff5/t7TuiAcIrL3JS4PsRIp+1/x15DXGzuHQehrtc5rMiEb1vFCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "VzaYwMVgqCqeX66cmd21g+qey9AXFL3q04dTRasJH0s="
          },
          "tokens_per_usd_hour": 55
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "IQ4_XS (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-iq4_xs-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_XS",
          "engine": "ollama-host",
          "tokens_per_sec": 6.93,
          "cost_usd_per_million": 4.7232,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "APPN3tO3Hy3RRDM5bpPvLeAadDpcGc5M/dGvzK0TtEb3Ji0Z/MRDwdsbfRWRgXkLn+00w4FThJc2tG4GfwJFBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "nuEAkZfrnuT21JxTIe560JywU1vX1SKGrJF0SwwTVyU="
          },
          "tokens_per_usd_hour": 59
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": "IQ1_S (1-bit) by maziyarpanahi",
          "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-iq1_s-ollama",
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:IQ1_S",
          "engine": "ollama-host",
          "tokens_per_sec": 6.6,
          "cost_usd_per_million": 4.6098,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "YkH8hjmB0D1RxTb0myTIO3rpNHcKDliF1WDro1t+hjnHFw97QCDFGw/BUbLpk7NZV43F1I7dAfklAXD0xaoZBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "l/x+eT5oT+zLWaIe2WjY8QxKTNUTZqNv7zx60d1ALWA="
          },
          "tokens_per_usd_hour": 60
        },
        {
          "model": "Hunyuan MT 2 30B-A3B",
          "page": "/inferences/hy-mt-2/30b-a3b",
          "build": "Q4_K_M (4-bit) by tencent",
          "build_page": "/inferences/hy-mt-2/30b-a3b/tencent-q4_k_m-ollama",
          "ref": "hf.co/tencent/Hy-MT2-30B-A3B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 5.36,
          "cost_usd_per_million": 8.3129,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "RTFgNEfiOHEhYFohoGoxFFyKell+w3kWlYqqiVggODJk9UopRoPwCwPSPfOtJV2qp4j5pZnjeRgHnIcpU0LeCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "MntXlcKKYxYiFJcxaBqL2uG+JQGUu5LAZo85uJSjD3Y="
          },
          "tokens_per_usd_hour": 33
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q8_0 (8-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q8_0-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 5.33,
          "cost_usd_per_million": 6.4387,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "lag0lwjS4kyWzwpRkRHVnEMNQH0ca4joB4rweunr9B64XjrGwSOY0rioLtiDVxRY3fIyXtD8bDGVeHp4MNKODg==",
            "signing_payload_b64": "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",
            "payload_sha256": "JiwOrHWqI9pNus/avHdZrof1LjcUpdDMap6yHsAhQDg="
          },
          "tokens_per_usd_hour": 43
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": "IQ3_M (3-bit) by bartowski",
          "build_page": "/inferences/llama-3-2/1b/bartowski-iq3_m-ollama",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:IQ3_M",
          "engine": "ollama-host",
          "tokens_per_sec": 5.28,
          "cost_usd_per_million": 5.862,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "bIEnv7GZNBLKxTlX0PehieTL3iLHvTLkO+GhNpz73QZ2inxjwS25vaLTi7Y0gutWwcBMsfOy4z+/KqRxHiKYAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "L4owj79V9xN0X3NttAyRjeSHHK01XGMNfVAWdbbO1ms="
          },
          "tokens_per_usd_hour": 47
        },
        {
          "model": "Kimi Linear 48B-A3B",
          "page": "/inferences/kimi-linear/48b-a3b",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/kimi-linear/48b-a3b/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/moonshotai_Kimi-Linear-48B-A3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 4.81,
          "cost_usd_per_million": 11.1806,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "YoELUsrRXI23U/VSXa434Rs3mzV4HdxRxj5kFzfivKCYDZLhtsLqnLZ1a8LlQsgSq1lCJ0CAZd82m9FAFNTfDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "Fus3Neuar8wsWRwwszugDZ3VoBe/dlQ3FA4A63ERgZo="
          },
          "tokens_per_usd_hour": 25
        },
        {
          "model": "Kimi-Linear-Reap-35B-A3B-Instruct",
          "page": "/inferences/kimi-linear/35b-a3b~cerebras-reap",
          "build": "Q4_K_M (4-bit) by mradermacher",
          "build_page": "/inferences/kimi-linear/35b-a3b~cerebras-reap/mradermacher-q4_k_m-ollama",
          "ref": "hf.co/mradermacher/Kimi-Linear-REAP-35B-A3B-Instruct-i1-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 4.75,
          "cost_usd_per_million": 9.8821,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "O50OXK3rPvrKCrLG8lS0x49NkRYMG3tasYGmx4OSND4NUQjsWXZzfn83hAm55qHamCDhPTZdsE11kS63DP+yAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "+cfCglecb3AakW6wGod4BVL5rY0/BWZJFkkBhNI+uck="
          },
          "tokens_per_usd_hour": 28
        },
        {
          "model": "Granite 4.0 1B",
          "page": "/inferences/granite-4-0/1b",
          "build": "BF16 (16-bit, full precision) by ibm-granite",
          "build_page": "/inferences/granite-4-0/1b/ibm-granite-bf16-ollama",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:BF16",
          "engine": "ollama-host",
          "tokens_per_sec": 4.71,
          "cost_usd_per_million": 6.9928,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "Q5MaNrs4bUmuY7+o48kTpDfNpEdtCywAvzEVRUl9Jadv2QI7VGC63p91qww8F5G48/MLZp64jov+tWCq82G4Dg==",
            "signing_payload_b64": "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",
            "payload_sha256": "CHwLS3E2726+k6/CPWzr4fHNlFBUt5+NBi/nBWqpHkM="
          },
          "tokens_per_usd_hour": 40
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 4.41,
          "cost_usd_per_million": 8.9491,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "+G+ybq6tBWwm8EeOJZCgzn79CShJvim8ErkVPpzvObHOUhDPdB/KA3o9E7uGP6DiI32t9OYZET9XE/9EhybnCA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NC40MDcwMTM2NDQ5NjYyODksInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzMzQyMTI0MDk2MDA0NzU4LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwYTQ4NGY4YTBhM2U0MjcyNmEzOGUxZjE1NzdmNWJlZTUzMGU1MTg1NzE4NGE4MWJhZTg5Njc2MzVkNmUzNjU0ZiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbm9uZS8zMmciLCJtb2RlbCI6ImhmLmNvL3Vuc2xvdGgvZ3B0LW9zcy0yMGItR0dVRjpRNF9LX00iLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6NDIsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6MTYwMC44NTczMDc5OTk5OTk2LCJ0b3RhbF90b2tlbnMiOjcwNTV9fQ==",
            "payload_sha256": "0R0oWGjp4U8rA980GTXB+sz8mG6Wr/4Ig/X4ubJEbIc="
          },
          "tokens_per_usd_hour": 31
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": "F16 (16-bit, full precision) by bartowski",
          "build_page": "/inferences/llama-3-2/1b/bartowski-f16-ollama",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:F16",
          "engine": "ollama-host",
          "tokens_per_sec": 4.3,
          "cost_usd_per_million": 7.5285,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "OcsybWnv9AQwqxC0kOi5jO9i+oxUXhvxsnlYtvieoO1uZ44LaWmSAWeSqkRqkySdAW7dkiv4iuSPN+ixO72MAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "/CbkYXkRopu7qgHrvn1xgFSmnOxl4/NX/aDquw//4AE="
          },
          "tokens_per_usd_hour": 37
        },
        {
          "model": "Qwen 3.5 35B-A3B",
          "page": "/inferences/qwen-3-5/35b-a3b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/qwen-3-5/35b-a3b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 3.9,
          "cost_usd_per_million": 12.1846,
          "measured_at": "2026-11-15T06:44:44.996Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "Du0/LK/kmEofPwPX5IBXRRXpL3ctP6QIaploNM7/Wsh1e+7Z5tTLuXPIjY/y2SOSCeor1+glC/2yGRDN0p7uCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "OB3L4nxoTRZGU3Cspqtrp+iqA55rIvR9fygLBvbt0Ts="
          },
          "tokens_per_usd_hour": 23
        },
        {
          "model": "LFM2.5 24B-A2B",
          "page": "/inferences/lfm-2-5/24b-a2b",
          "build": "IQ3_XXS (3-bit) by unsloth",
          "build_page": "/inferences/lfm-2-5/24b-a2b/unsloth-iq3_xxs-ollama",
          "ref": "hf.co/unsloth/LFM2.5-24B-A2B-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "tokens_per_sec": 3.64,
          "cost_usd_per_million": 10.2663,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "tYkrlLfxdoF3XBFrQITD54Z8TsIKIZ6w7cOJ/Rf7ozaSpSNvJ9xHImlFCdxXTaK/Vb/rqrcOA2O/uN5RWPlMAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "asYA5ffSMT9zb0bo3p3aQSrFWGP33B0S58oc8IPopfA="
          },
          "tokens_per_usd_hour": 27
        },
        {
          "model": "GLM-4.7 Flash",
          "page": "/inferences/glm-4-7/flash",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/glm-4-7/flash/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/GLM-4.7-Flash-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 3.55,
          "cost_usd_per_million": 12.5701,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "xGfHrry32UfdIWxUCLLWOfeG3HgDHy04wN3bxNtJbQAlnvDAGvkMyU7Ylr7eRz0kwhHxyVWo93aRzVURcEogBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "M+qXLEAZFbxdEpXxUK7hawQRcB1vxQyNmAGjYWGeEf0="
          },
          "tokens_per_usd_hour": 22
        },
        {
          "model": "Qwen3.8-27B-Uncensored-GGUF",
          "page": "/inferences/qwen-3-8/27b~jonathancoletti-uncensored",
          "build": "Q5_K_M (5-bit) by JonathanColetti",
          "build_page": "/inferences/qwen-3-8/27b~jonathancoletti-uncensored/jonathancoletti-q5_k_m-ollama",
          "ref": "hf.co/JonathanColetti/Qwen3.8-27B-Uncensored-GGUF:Q5_K_M",
          "engine": "ollama",
          "tokens_per_sec": 3.37,
          "cost_usd_per_million": null,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "AL5e4Zc/CHAs2xEAIF2X+oVXoRzay10SMnbcme5quL/5kJZeS6o/k/eoHsTxTlIcM18q9YSFlereiMRCWSLVBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "Uu9bCdn9pUcPbGtiZOQLolmNDP5lIHjFNM8j35mjzfE="
          },
          "tokens_per_usd_hour": null
        },
        {
          "model": "North 1.0 Mini-Code-1-0",
          "page": "/inferences/north-1/mini-code-1-0",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/north-1/mini-code-1-0/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/North-Mini-Code-1.0-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 3.2,
          "cost_usd_per_million": 14.1577,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "4NB1Ukny6rhc1gY5u7446P3vFgG6oB9moRtcbcbWfww4UJOlNqUWfOT/66qELlEB+cwYmROmGNEnDD8wVrq1Cw==",
            "signing_payload_b64": "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",
            "payload_sha256": "RUJWzDwzT1IdZg+m2HmpV1AKoMLpujT0k3JPev+qVZc="
          },
          "tokens_per_usd_hour": 20
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "BF16 (16-bit, full precision) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-bf16-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:BF16",
          "engine": "ollama-host",
          "tokens_per_sec": 3.14,
          "cost_usd_per_million": 11.9727,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "R55I6Zr0hGc5e3kdiZi4j5CevKAvxSkvg3dMzkXrrnvrBMXcRB8e/nVe6GMymbJUmV3ZZd+PUp4xDjc0JXuOCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "++4DHPgbc72KmX2S0OxSXVVxNb4R159M4IXYxqMSBC4="
          },
          "tokens_per_usd_hour": 23
        },
        {
          "model": "Qwen 3.5 9B",
          "page": "/inferences/qwen-3-5/9b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/qwen-3-5/9b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Qwen3.5-9B-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 2.39,
          "cost_usd_per_million": null,
          "measured_at": "2026-10-30T04:00:04.405Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "hW44MIAlhpIPtlh9xuVlT/Dpy0a0MTg5/IgLpjdl8QkdYc3DNhGfreArnIAUjTVTFDj119t4qsBiLeusr8oZDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "oz8IG1hK8gawyLDYnAqOzN4JM0bBpyglTJi6n/FlKFc="
          },
          "tokens_per_usd_hour": null
        },
        {
          "model": "Granite 4.2 8B",
          "page": "/inferences/granite-4-2/8b",
          "build": "Q4_K_M (4-bit) by ibm-granite",
          "build_page": "/inferences/granite-4-2/8b/ibm-granite-q4_k_m-ollama",
          "ref": "hf.co/ibm-granite/granite-4.2-8b-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 1.87,
          "cost_usd_per_million": null,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "icyyvBZSBH8lH1/TgpSefrdPBjZGm7uw9HoF1d9lnZDlW29iDNxxyYJRj+QTBCuiLdOWLhXhTATNvxPWyvHjDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "tcKInU+NfUeuEfNy92D6azixWacAx5MnDObyFi0Avdo="
          },
          "tokens_per_usd_hour": null
        },
        {
          "model": "Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP-GGUF",
          "page": "/inferences/qwen-3-8/27b~hauhaucs-uncensored-aggressive-mtp",
          "build": "Q6_K_P (6-bit) by HauhauCS",
          "build_page": "/inferences/qwen-3-8/27b~hauhaucs-uncensored-aggressive-mtp/hauhaucs-q6_k_p-ollama",
          "ref": "hf.co/HauhauCS/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP-GGUF:Q6_K_P",
          "engine": "ollama",
          "tokens_per_sec": 1.75,
          "cost_usd_per_million": null,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 65536,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "4N4JLpP8Ih/qE3Dp3fyNdxdDgN/1G5vp/AL8Is22PHkX842BuutWc77Hx2iNZNsLLAHS11kYTW49kinSNVUzCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "XgyhbxXqAxhc034fvYmVMaFK6+b7s2ThQeN4nlWPMg0="
          },
          "tokens_per_usd_hour": null
        }
      ],
      "rank": 2,
      "of": 36,
      "percentiles": {
        "gpu_tflops": 0,
        "memory_bandwidth": 63,
        "cpu_all_cores": 36,
        "cpu_one_core": 32,
        "disk_read": 75,
        "download": 50
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "dae679",
      "did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
      "name": "Gram ·dae679",
      "claimed": true,
      "online": false,
      "last_seen": "2026-11-07T05:15:02.871Z",
      "version": "a5f43c7",
      "region": {
        "code": "SG_190464",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/nvidia/64g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2.2,
        "cpus": 12,
        "ram_gb": 52.95807647705078,
        "memory_bandwidth_gb_s": 2.8,
        "gpu": "NVIDIA L4",
        "vram_gb": 22,
        "gpu_tflops_fp16": 52.7,
        "disk_read_mb_s": 3356.3,
        "gpu_count": 1,
        "gpu_org": "nvidia",
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
          "clock_ghz": 2.2,
          "cores_logical": 12,
          "cores_physical": 6,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 1024,
          "l3_mb": 38.5,
          "single_core_hashes_per_sec": 295.2,
          "multi_core_hashes_per_sec": 1700.6,
          "sustained_ratio": 0.9,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-07T04:46:20.920Z",
          "measured_in": "SG_190464",
          "signed": true
        },
        "memory": {
          "total_gb": 53,
          "bandwidth_gb_s": 2.8,
          "latency_ns": 23.6,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-07T04:46:21.530Z",
          "measured_in": "SG_190464",
          "signed": true
        },
        "gpu": {
          "name": "NVIDIA L4",
          "memory_gb": 22,
          "fp16_tflops": 52.7,
          "watts": 34.2,
          "tflops_per_watt": 1.542,
          "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
          "measured_at": "2026-11-07T04:46:37.158Z",
          "measured_in": "SG_190464",
          "signed": true
        },
        "storage": {
          "read_mb_s": 3356.3,
          "write_mb_s": 287.7,
          "random_read_iops": 812404,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-11-07T04:46:22.036Z",
          "measured_in": "SG_190464",
          "signed": true
        }
      },
      "signed_runs": 76,
      "signed_runs_30d": 36,
      "builds_measured": 76,
      "best_results": 35,
      "first_signed_at": "2026-11-03T19:07:50.085Z",
      "last_signed_at": "2026-11-07T05:09:52.971Z",
      "awake_pct": 100,
      "queue": {
        "measured": 68,
        "failed": 15,
        "queued": 1835
      },
      "download_mib_per_sec": 116.8,
      "uptime_days": [
        {
          "date": "2026-11-03",
          "hours": 6
        },
        {
          "date": "2026-11-06",
          "hours": 19
        },
        {
          "date": "2026-11-07",
          "hours": 6
        }
      ],
      "uptime_pct": 43.1,
      "best_value": {
        "model": "FunctionGemma 270M",
        "page": "/inferences/functiongemma/270m",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/functiongemma/270m/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 340.82,
        "cost_usd_per_million": 0.6334,
        "measured_at": "2026-11-07T05:09:52.970Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "VCdNM5VrOGUuczk4NN28xP8Nl9jB+1kfCsrtkodSKIUketpEGf2CEpEV/i6bXq4lzOPVswFvfzsM3er1jegYBQ==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MzQwLjgyMTM0MzM5MTkxOTg3LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5NDAyODE5Mjk3MDk3MjIwNSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDUyNDA1MGU5Mjk2ZDMxMmNkMjgyOGM0MjYxNjViZmYyM2JhZWVmNzY5YWRhOTcwODMwOTg4MzJmMDZkYWU2NzkiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L252aWRpYS82NGciLCJtb2RlbCI6ImhmLmNvL3Vuc2xvdGgvZnVuY3Rpb25nZW1tYS0yNzBtLWl0LUdHVUY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjQ5LCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjQ0LjM2OTI4Njk5Nzk5OTk5LCJ0b3RhbF90b2tlbnMiOjE1MTIyfX0=",
          "payload_sha256": "W3nVeObvEQ2mzicETEDWcLKx7EWwJKsK5qdvjaJRh9k="
        },
        "tokens_per_usd_hour": 439
      },
      "opened": [
        {
          "ref": "hf.co/openbmb/MiniCPM5-1B-GGUF:Q4_K_M",
          "family": "minicpm-5",
          "member": "1b",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_M",
          "family": "hy-mt-2",
          "member": "1.8b",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/openbmb/MiniCPM-V-4.6-gguf:Q4_K_M",
          "family": "minicpm-v-4-6",
          "member": "base",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_M",
          "family": "deepseek-ocr-2",
          "member": "base",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "family": "llama-3-2",
          "member": "3b",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/bullerwins/translategemma-4b-it-GGUF:Q4_K_M",
          "family": "translategemma",
          "member": "4b",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q4_K_M",
          "family": "glm-ocr",
          "member": "base",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q8_0",
          "family": "minicpm-v-4-7",
          "member": "1b",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/sabafallah/DeepSeek-OCR-GGUF:Q4_K_M",
          "family": "deepseek-ocr",
          "member": "base",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/tencent/HY-MT1.5-1.8B-GGUF:Q4_K_M",
          "family": "hy-mt-1-5",
          "member": "1.8b",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3.5-4B-GGUF:Q4_K_M",
          "family": "qwen-3-5",
          "member": "4b",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/bartowski/HuggingFaceTB_SmolLM3-3B-GGUF:Q4_K_M",
          "family": "smollm-3",
          "member": "3b",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/mradermacher/MiniMax-M2.7-abliterated-BF16-i1-GGUF:IQ1_S",
          "family": "minimax-m2-7",
          "member": "abliterated~youssofal",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/openbmb/MiniCPM-o-4_5-gguf:Q4_K_M",
          "family": "minicpm-o-4-5",
          "member": "base",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/openbmb/MiniCPM-V-4_5-gguf:Q4_K_M",
          "family": "minicpm-v-4-5",
          "member": "base",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q4_K_M",
          "family": "minicpm-v-4-7",
          "member": "1b",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/openbmb/MiniCPM4.1-8B-GGUF:Q4_K_M",
          "family": "minicpm-4-1",
          "member": "8b",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/Reza2kn/Qianfan-OCR-GGUF:Q4_K_M",
          "family": "qianfan-ocr",
          "member": "base",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/unsloth/granite-4.1-3b-GGUF:Q4_K_M",
          "family": "granite-4-1",
          "member": "3b",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/unsloth/Magistral-Small-2506-GGUF:Q4_K_M",
          "family": "magistral-small",
          "member": "2506",
          "at": "2026-11-07T05:09:52.971Z"
        },
        {
          "ref": "hf.co/unsloth/medgemma-1.5-4b-it-GGUF:Q4_K_M",
          "family": "medgemma-1-5",
          "member": "4b",
          "at": "2026-11-07T05:09:52.971Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-11-03",
          "runs": 47
        },
        {
          "date": "2026-11-06",
          "runs": 357
        },
        {
          "date": "2026-11-07",
          "runs": 420
        }
      ],
      "runs_by_family": [
        {
          "family": "whisper",
          "runs": 4
        },
        {
          "family": "sd-turbo",
          "runs": 3
        },
        {
          "family": "glm-ocr",
          "runs": 3
        },
        {
          "family": "gemma-4",
          "runs": 2
        },
        {
          "family": "musicgen",
          "runs": 2
        },
        {
          "family": "depth-anything",
          "runs": 2
        },
        {
          "family": "nllb",
          "runs": 2
        },
        {
          "family": "moonshine",
          "runs": 2
        },
        {
          "family": "f5-tts",
          "runs": 2
        },
        {
          "family": "sana",
          "runs": 2
        },
        {
          "family": "hy-mt-1-5",
          "runs": 2
        },
        {
          "family": "minicpm-v-4-7",
          "runs": 2
        },
        {
          "family": "glm-4-7",
          "runs": 1
        },
        {
          "family": "sam",
          "runs": 1
        },
        {
          "family": "sarvam",
          "runs": 1
        },
        {
          "family": "tiny-sd",
          "runs": 1
        },
        {
          "family": "gemma-3",
          "runs": 1
        },
        {
          "family": "gpt-oss",
          "runs": 1
        },
        {
          "family": "kokoro",
          "runs": 1
        },
        {
          "family": "mms-tts",
          "runs": 1
        },
        {
          "family": "phi-4",
          "runs": 1
        },
        {
          "family": "minicpm-5",
          "runs": 1
        },
        {
          "family": "embeddinggemma",
          "runs": 1
        },
        {
          "family": "hy-mt-2",
          "runs": 1
        },
        {
          "family": "qwen-3-6",
          "runs": 1
        },
        {
          "family": "cogvideox",
          "runs": 1
        },
        {
          "family": "seamless-m4t",
          "runs": 1
        },
        {
          "family": "krutrim",
          "runs": 1
        },
        {
          "family": "granite-4-2",
          "runs": 1
        },
        {
          "family": "muse-glimmer",
          "runs": 1
        },
        {
          "family": "ministral-3",
          "runs": 1
        },
        {
          "family": "airavata",
          "runs": 1
        },
        {
          "family": "minicpm-v-4",
          "runs": 1
        },
        {
          "family": "minicpm-v-4-6",
          "runs": 1
        },
        {
          "family": "qwen-3",
          "runs": 1
        },
        {
          "family": "deepseek-ocr-2",
          "runs": 1
        },
        {
          "family": "functiongemma",
          "runs": 1
        },
        {
          "family": "llama-3-2",
          "runs": 1
        },
        {
          "family": "wespeaker",
          "runs": 1
        },
        {
          "family": "ecapa-tdnn",
          "runs": 1
        },
        {
          "family": "madlad",
          "runs": 1
        },
        {
          "family": "translategemma",
          "runs": 1
        },
        {
          "family": "openhathi",
          "runs": 1
        },
        {
          "family": "deepseek-ocr",
          "runs": 1
        },
        {
          "family": "qwen-3-5",
          "runs": 1
        },
        {
          "family": "sarvam-translate",
          "runs": 1
        },
        {
          "family": "qwen-3-8",
          "runs": 1
        },
        {
          "family": "xtts",
          "runs": 1
        },
        {
          "family": "smollm-3",
          "runs": 1
        },
        {
          "family": "minimax-m2-7",
          "runs": 1
        },
        {
          "family": "embeddinggemma-2",
          "runs": 1
        },
        {
          "family": "minicpm-o-4-5",
          "runs": 1
        },
        {
          "family": "minicpm-v-4-5",
          "runs": 1
        },
        {
          "family": "minicpm-4-1",
          "runs": 1
        },
        {
          "family": "qianfan-ocr",
          "runs": 1
        },
        {
          "family": "granite-4-1",
          "runs": 1
        },
        {
          "family": "magistral-small",
          "runs": 1
        },
        {
          "family": "medgemma-1-5",
          "runs": 1
        },
        {
          "family": "glm-4-5",
          "runs": 1
        },
        {
          "family": "hy-3",
          "runs": 1
        }
      ],
      "rooms": [
        "SG_190464"
      ],
      "awards": [
        {
          "id": "runs-10",
          "label": "10+ signed runs"
        },
        {
          "id": "opener",
          "label": "First to measure 21 builds"
        },
        {
          "id": "best-10",
          "label": "Best result on 35 builds"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "FunctionGemma 270M",
        "page": "/inferences/functiongemma/270m",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/functiongemma/270m/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 340.82,
        "cost_usd_per_million": 0.6334,
        "measured_at": "2026-11-07T05:09:52.970Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "VCdNM5VrOGUuczk4NN28xP8Nl9jB+1kfCsrtkodSKIUketpEGf2CEpEV/i6bXq4lzOPVswFvfzsM3er1jegYBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "W3nVeObvEQ2mzicETEDWcLKx7EWwJKsK5qdvjaJRh9k="
        },
        "tokens_per_usd_hour": 439
      },
      "cheapest": {
        "model": "FunctionGemma 270M",
        "page": "/inferences/functiongemma/270m",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/functiongemma/270m/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 340.82,
        "cost_usd_per_million": 0.6334,
        "measured_at": "2026-11-07T05:09:52.970Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "VCdNM5VrOGUuczk4NN28xP8Nl9jB+1kfCsrtkodSKIUketpEGf2CEpEV/i6bXq4lzOPVswFvfzsM3er1jegYBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "W3nVeObvEQ2mzicETEDWcLKx7EWwJKsK5qdvjaJRh9k="
        },
        "tokens_per_usd_hour": 439
      },
      "builds": [
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/functiongemma/270m/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 340.82,
          "cost_usd_per_million": 0.6334,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "VCdNM5VrOGUuczk4NN28xP8Nl9jB+1kfCsrtkodSKIUketpEGf2CEpEV/i6bXq4lzOPVswFvfzsM3er1jegYBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "W3nVeObvEQ2mzicETEDWcLKx7EWwJKsK5qdvjaJRh9k="
          },
          "tokens_per_usd_hour": 439
        },
        {
          "model": "DeepSeek-OCR Base",
          "page": "/inferences/deepseek-ocr/base",
          "build": "Q4_K_M (4-bit) by sabafallah",
          "build_page": "/inferences/deepseek-ocr/base/sabafallah-q4_k_m-ollama",
          "ref": "hf.co/sabafallah/DeepSeek-OCR-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 340.22,
          "cost_usd_per_million": 0.6384,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "MldOjzjP4hwZYEELAoIg6JxlNEbRpD5Jl76MA5awvmFEE1hThq8QYNnXKQkzyGga54Xr+8vBwD+lps1pWwABCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "jHQ1GDVSNrtTn5osqOMTcXALb6OfiZocbGwYFpH6n2w="
          },
          "tokens_per_usd_hour": 435
        },
        {
          "model": "DeepSeek-OCR 2 Base",
          "page": "/inferences/deepseek-ocr-2/base",
          "build": "Q4_K_M (4-bit) by sabafallah",
          "build_page": "/inferences/deepseek-ocr-2/base/sabafallah-q4_k_m-ollama",
          "ref": "hf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 334.82,
          "cost_usd_per_million": 0.6487,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "dnan915zASGOA/HXQ3Ttqx4niiHeRKYwoxlgAdV3bu9/RxWsak30C6bwLkw22mpr84jSJpbvh4ws6/KcHvRbCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "At5RWBQFLG5B6sgKkhEe1ty7aQJ7BhyiNckdD899ZwQ="
          },
          "tokens_per_usd_hour": 428
        },
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 291.82,
          "cost_usd_per_million": 0.7398,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "Zpk8DwNmxZpWQAp8UvvR1cs3ESn7OOqBA9uvazjfGikH8GDXupDYQQVssZmtHBPLyhbjtrCDSWbaitPrrebODw==",
            "signing_payload_b64": "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",
            "payload_sha256": "bVI8hPE20wRzMSYiO3EsoHBaoPDwzU6QOO4AUr9wyrQ="
          },
          "tokens_per_usd_hour": 375
        },
        {
          "model": "MiniCPM-V 4.7 1B",
          "page": "/inferences/minicpm-v-4-7/1b",
          "build": "Q4_K_M (4-bit) by openbmb",
          "build_page": "/inferences/minicpm-v-4-7/1b/openbmb-q4_k_m-ollama",
          "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 221.15,
          "cost_usd_per_million": 0.9777,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 262144,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "nt3XDnfvCaYE1DSM78DqccE1gTM/Inf1U1d3NVrcFbMdzA5EFUPa1yxfBPNoeYQbsGEwEDxA3hcbDM2Qw54BAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "OVOtN858raplNZexGPF/dG6nWWbKvgPPlaz5RItKc54="
          },
          "tokens_per_usd_hour": 284
        },
        {
          "model": "MiniCPM5 1B",
          "page": "/inferences/minicpm-5/1b",
          "build": "Q4_K_M (4-bit) by openbmb",
          "build_page": "/inferences/minicpm-5/1b/openbmb-q4_k_m-ollama",
          "ref": "hf.co/openbmb/MiniCPM5-1B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 208.3,
          "cost_usd_per_million": 1.0381,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 65536,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "Oqg8YQNW3f1j701J0bwXWiEJMF3KbfvCKE9GgZ5FtkknN1I52uQUgZA9cfr+3MZS8Je6SEF9vYdbntxOxkJDAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "xboKOF4UlUmAdXUmNXxRHX1CuzgXDXoyabPHZo1QVww="
          },
          "tokens_per_usd_hour": 268
        },
        {
          "model": "MiniCPM-V 4.7 1B",
          "page": "/inferences/minicpm-v-4-7/1b",
          "build": "Q8_0 (8-bit) by openbmb",
          "build_page": "/inferences/minicpm-v-4-7/1b/openbmb-q8_0-ollama",
          "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 185.86,
          "cost_usd_per_million": 1.1651,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 262144,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "tGG/WQL7bWmpumRYSXrjkTCZH6N+pMvcOaoM0ZngBAUDQBeEZf93nKsEL/ELJOAME+TVJqCjhR7wu3xyNuBPAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "XfQVj5GMlUEiWg+ouAjWE29Nu8Nb2J8UIPiY3+hZ7UE="
          },
          "tokens_per_usd_hour": 238
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by google",
          "build_page": "/inferences/gemma-4/e2b/google-q4_k_m-ollama",
          "ref": "gemma4:e2b",
          "engine": "ollama-host",
          "tokens_per_sec": 149.89,
          "cost_usd_per_million": 1.4779,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "geBl5+OCuwaouiLNj4hTEXccZoUTUhxy8pspAzA0cbYywlce34JGlhJmlnjnSjjQld3udO9AClLy6gn+cht/CA==",
            "signing_payload_b64": "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",
            "payload_sha256": "KaXNHyevfmFklL4IbToF97deO3IIeXBfknigo3L9lmk="
          },
          "tokens_per_usd_hour": 188
        },
        {
          "model": "MiniCPM-V 4.6 Base",
          "page": "/inferences/minicpm-v-4-6/base",
          "build": "Q4_K_M (4-bit) by openbmb",
          "build_page": "/inferences/minicpm-v-4-6/base/openbmb-q4_k_m-ollama",
          "ref": "hf.co/openbmb/MiniCPM-V-4.6-gguf:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 130.91,
          "cost_usd_per_million": 1.6508,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 262144,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "yk9bevepOPKKjbVBVlsSlq4Jx5v0/2eP+Sx4OB43cug8U6GZ+3nlK40HK8jCrqoP91Fr+LIARO9kqVy2C4pFDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "rmI44TzVINHmIPYTKm1EB2JYyrW2u5M1fRde5MAMkWA="
          },
          "tokens_per_usd_hour": 168
        },
        {
          "model": "Sarvam-1",
          "page": "/inferences/sarvam/1",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 121.44,
          "cost_usd_per_million": 1.7861,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "RDQRzBO3Ke2zwZxG59t3qpuu4vt+0iyh+KT5WvNfvzFgNA7u9czHmPcg+QXP9UQD2DuW4s9XkZB1Olzfa3cuCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "kKojnbs6fubM5IETsHxYSbTKvI0PpXNM+4MHaHbzCDI="
          },
          "tokens_per_usd_hour": 156
        },
        {
          "model": "GLM-OCR Base",
          "page": "/inferences/glm-ocr/base",
          "build": "Q4_K_M (4-bit) by mradermacher",
          "build_page": "/inferences/glm-ocr/base/mradermacher-q4_k_m-ollama",
          "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 106.1,
          "cost_usd_per_million": 2.037,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "V/tN/jFjOUBl+pYQsVxuIr+guLrtf7ineA8t0YZRhoiqFy2pfa+VQvCs/oEFBkBygoaEb53lY4MZGjyTbYQEBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "RykXd/aXfdqnUkhOUdeFhTgfrHWKtLyUK/AZUO/OSTY="
          },
          "tokens_per_usd_hour": 136
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "tokens_per_sec": 98.85,
          "cost_usd_per_million": 2.2141,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "0ojOPVeq3RHALFzo3fjit9ENkN5A6Kg+7QvwzadHx2LbjjPQGmU7IRgYm36SobK4PYQapZua3fjyupR/4J+tAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "629GRJoEP+hRBaykSjsTqOifOWURQv1eot/WIwrB+qs="
          },
          "tokens_per_usd_hour": 125
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/llama-3-2/3b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 97.41,
          "cost_usd_per_million": 2.2303,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "UUp1dqpyvr1/PNTEE47OUMP7x4DpkyQZwkT0rpFYMZLlDpxlJ97SNXGq37Lz8dUn0a/r8YgZZ3BB3Q6HKA/jBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "e+QldB5nxZZXzSNtgbc2pv7T4q7lF1tzpI4QKaybOjA="
          },
          "tokens_per_usd_hour": 125
        },
        {
          "model": "MiniCPM-V 4 Base",
          "page": "/inferences/minicpm-v-4/base",
          "build": "Q4_K_M (4-bit) by openbmb",
          "build_page": "/inferences/minicpm-v-4/base/openbmb-q4_k_m-ollama",
          "ref": "hf.co/openbmb/MiniCPM-V-4-gguf:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 96.71,
          "cost_usd_per_million": 2.2478,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "1zm49qNy9FzO+ULoa6Xue2ZXqidyUIT8SRPzUXTgxVSTuxOglEkfOSenUcSJQ0gz0CEi9DBVrQLLSaNJGe8/BA==",
            "signing_payload_b64": "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",
            "payload_sha256": "YpGidNliB60trWm3kXZb/FY3+KBbBNf+yhuPj+vM448="
          },
          "tokens_per_usd_hour": 124
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 79.36,
          "cost_usd_per_million": 2.7424,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "ZVrH7Uw3fgyUscrLT6EXcr8fmS3oA1+N2BRHjNs/ka2uCbqy6FNUJEtXa3kzJaFVKwDxZREZzhGcBqx97OMsCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "6XmtTI7MMrj3CZInM9+gejn4BOVgLv/ws5x5QXv+o+U="
          },
          "tokens_per_usd_hour": 101
        },
        {
          "model": "SmolLM3 3B",
          "page": "/inferences/smollm-3/3b",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/smollm-3/3b/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/HuggingFaceTB_SmolLM3-3B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 77.59,
          "cost_usd_per_million": 2.7991,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 65536,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "vNQ+jdq5W0Uyi9s2bF8+JB4DGH/s8siZsoIGHtjGso1zJKQMcC4dyzfKzy36y92nHJaZq+zRdAn9b/y9LHYABg==",
            "signing_payload_b64": "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",
            "payload_sha256": "y9ujbzKC/wGIGaYxwDx9Qi6PfdsibgyamJTLtqree54="
          },
          "tokens_per_usd_hour": 99
        },
        {
          "model": "TranslateGemma 4B",
          "page": "/inferences/translategemma/4b",
          "build": "Q4_K_M (4-bit) by bullerwins",
          "build_page": "/inferences/translategemma/4b/bullerwins-q4_k_m-ollama",
          "ref": "hf.co/bullerwins/translategemma-4b-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 77.54,
          "cost_usd_per_million": 2.8066,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "cKhWjOmRDh9Gk2zmUbPpAeuHGjrYWQ2myibKuk9lMVKWOfqbwFL3iyjhQNMNUFD3DY23llJgdWht7dAMumQtBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "qHZIk/D5bwnu2sLwRk+knlumQVrQ3izZO9C47Yt6GUA="
          },
          "tokens_per_usd_hour": 99
        },
        {
          "model": "MedGemma 1.5 4B",
          "page": "/inferences/medgemma-1-5/4b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/medgemma-1-5/4b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/medgemma-1.5-4b-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 74.52,
          "cost_usd_per_million": 2.9204,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 65536,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "ZsHHtynPKIw3Fmzuq6EyM1UOiq5yUj9LuKWRsrqwyghaUiFGR7CxdoeYeDpYc8e0V7IsrAzhr2gAjH1jSpqYAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "GSsNT9GperqViKgHQMXM9W0aZvaPrFD+wUwOGznNgbQ="
          },
          "tokens_per_usd_hour": 95
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "Q4_K_M (4-bit) by Qwen",
          "build_page": "/inferences/qwen-3/4b/qwen-q4_k_m-ollama",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 49.84,
          "cost_usd_per_million": 4.3667,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 40960,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "1mCeWIXdawe2YkKNQXshmJtrsyuAk6FvaAaHkF+VVwWSE4GNfxaWHqkgqVVPs5ugisprL/wL0Ck3m7+z6kJkCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "ZS/WDMZMiN5S7xxwZSpynAE4YnfalyxMBIfEHwRJaRQ="
          },
          "tokens_per_usd_hour": 64
        },
        {
          "model": "Hunyuan MT 1.5 1.8B",
          "page": "/inferences/hy-mt-1-5/1.8b",
          "build": "Q4_K_M (4-bit) by tencent",
          "build_page": "/inferences/hy-mt-1-5/1.8b/tencent-q4_k_m-ollama",
          "ref": "hf.co/tencent/HY-MT1.5-1.8B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 46.49,
          "cost_usd_per_million": 4.6588,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "NNbTmCLwHGAorvh1JM04OX7kNiqaUDV9mpXoUKND7kbAu92PwN3OX5bpfGDlDetBsra/lb1h3y0u8tDqLxjeDA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NDYuNDg2NzYxMzgyNzY2NTYsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0MDI4MTkyOTcwOTcyMjA1LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwNTI0MDUwZTkyOTZkMzEyY2QyODI4YzQyNjE2NWJmZjIzYmFlZWY3NjlhZGE5NzA4MzA5ODgzMmYwNmRhZTY3OSIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbnZpZGlhLzY0ZyIsIm1vZGVsIjoiaGYuY28vdGVuY2VudC9IWS1NVDEuNS0xLjhCLUdHVUY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjUyLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjMwLjYzMjM3Njk5NjAwMDAxLCJ0b3RhbF90b2tlbnMiOjE0MjR9fQ==",
            "payload_sha256": "i6nuKjoVXCDaeuyXdIRf9fTeYHlJPwAxundLabnF1XY="
          },
          "tokens_per_usd_hour": 60
        },
        {
          "model": "MiniCPM4.1 8B",
          "page": "/inferences/minicpm-4-1/8b",
          "build": "Q4_K_M (4-bit) by openbmb",
          "build_page": "/inferences/minicpm-4-1/8b/openbmb-q4_k_m-ollama",
          "ref": "hf.co/openbmb/MiniCPM4.1-8B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 45.81,
          "cost_usd_per_million": 4.7923,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 65536,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "oKC0Iux/EPIFdOmaybf/XLlvTxWIFqlW5yz2G9gKZu14QMOxJBpUFBk3a/k7h9Rkl1sqdLvs3x06RO6AriNWBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "OKHWYI52qLuV/meKsWPrlTFLbpDvo/lrETS1IAnIJYE="
          },
          "tokens_per_usd_hour": 58
        },
        {
          "model": "Qwen 3.5 4B",
          "page": "/inferences/qwen-3-5/4b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/qwen-3-5/4b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Qwen3.5-4B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 41.82,
          "cost_usd_per_million": 5.209,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 262144,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "93zGjNlrQWFWFFfbgbTtz8mrqrntDDMrC+k7Aw+iSwdFVBuCZm1PUnuDezqXz9ogdPJLvOoe2EyPCBY7KYsYDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "GtCHeNIu4MOolLC6eTFXMdSUR9f3BcyoMcaR+VeO4mA="
          },
          "tokens_per_usd_hour": 53
        },
        {
          "model": "GLM-4.7 Flash",
          "page": "/inferences/glm-4-7/flash",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/glm-4-7/flash/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/GLM-4.7-Flash-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 40.58,
          "cost_usd_per_million": 5.6652,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "qOoNAso8oxKI5fA5h/+97Jp0fx1BfG9SK1selBw9lt6PEkaRaQCIlzcFIvKaA0KjcVkBZ55MVjX1mVL/yUifAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "oVORjB2am+l7Mp5SaeLxQ+bEdxG6eFMMo1+WE0OpX/Q="
          },
          "tokens_per_usd_hour": 49
        },
        {
          "model": "Hunyuan MT 2 1.8B",
          "page": "/inferences/hy-mt-2/1.8b",
          "build": "Q4_K_M (4-bit) by tencent",
          "build_page": "/inferences/hy-mt-2/1.8b/tencent-q4_k_m-ollama",
          "ref": "hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 36.5,
          "cost_usd_per_million": 5.9337,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "yTk/m+YOLv42J+26aZq/6M0CjCc/pO0gbVKFx5wvGlslvYMo8lgnDFEChtwwjHqrGO918YCfIzOELKW4vNObAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "99PE4dNrYcS6W3TTiS67PWoe2oOgZEF1NgRU6Ui7+r4="
          },
          "tokens_per_usd_hour": 47
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 24.03,
          "cost_usd_per_million": 9.3514,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "rzy7bhjq6GLr+VmK5W2qAfcs7X9nEp4dC6HMepoA2QgEdvS47oX1SkP/92P+Ca4bjy09aCwmBG+cxDKIcyWhCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "fV8qxbnMEuLlnWiKrTVxCNICsRJi1opdVXqscaeX924="
          },
          "tokens_per_usd_hour": 30
        },
        {
          "model": "MiniCPM-o 4.5 Base",
          "page": "/inferences/minicpm-o-4-5/base",
          "build": "Q4_K_M (4-bit) by openbmb",
          "build_page": "/inferences/minicpm-o-4-5/base/openbmb-q4_k_m-ollama",
          "ref": "hf.co/openbmb/MiniCPM-o-4_5-gguf:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 20.23,
          "cost_usd_per_million": 10.8538,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "ogwChQwpPuTIlu+sxQDjLCkJYV5a++0D44GKTGOGJSQT0M+Y8l0xDk2TuwyVBA9YSDZZcISLjUjVuC1zWYpzAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "E3L5XMdxEzr/2VHf6aD/uH9RVokArOO2o/yAEAh0vBQ="
          },
          "tokens_per_usd_hour": 26
        },
        {
          "model": "Granite 4.1 3B",
          "page": "/inferences/granite-4-1/3b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/granite-4-1/3b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/granite-4.1-3b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 18.15,
          "cost_usd_per_million": 11.9748,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 65536,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "Kks22pwqcoELK1LJREfFD+zJ3kmp/syvbvOSE1iHEUlb3XL/3DgLGzgAouogu5D6khef2xFTWvrvCk5k63CEAw==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTguMTQ4NDA2NTk0Mjk5NTA1LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5NDAyODE5Mjk3MDk3MjIwNSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDUyNDA1MGU5Mjk2ZDMxMmNkMjgyOGM0MjYxNjViZmYyM2JhZWVmNzY5YWRhOTcwODMwOTg4MzJmMDZkYWU2NzkiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L252aWRpYS82NGciLCJtb2RlbCI6ImhmLmNvL3Vuc2xvdGgvZ3Jhbml0ZS00LjEtM2ItR0dVRjpRNF9LX00iLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6NzAsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6MTIzLjcwMjMxOTk5OTAwMDA1LCJ0b3RhbF90b2tlbnMiOjIyNDV9fQ==",
            "payload_sha256": "Nmi0mfb/kuXpYt4fsQv9/mlpRMKoACofADmppw0PhlY="
          },
          "tokens_per_usd_hour": 23
        },
        {
          "model": "Qianfan-OCR Base",
          "page": "/inferences/qianfan-ocr/base",
          "build": "Q4_K_M (4-bit) by Reza2kn",
          "build_page": "/inferences/qianfan-ocr/base/reza2kn-q4_k_m-ollama",
          "ref": "hf.co/Reza2kn/Qianfan-OCR-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 17.56,
          "cost_usd_per_million": 12.4008,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "94bsoxIscQhgm/aSN+mpOeTO8FAXCcTcZos1xpksWmyNS7JSJQAUaURyZxxrExmvwKhvasBJ7Zb1IZYkqhELCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "ZljfoSOoxrrfF2AtTC3UHMPyaKjFZXsv6RPkq1WLO5k="
          },
          "tokens_per_usd_hour": 22
        },
        {
          "model": "Granite 4.2 3B",
          "page": "/inferences/granite-4-2/3b",
          "build": "Q4_K_M (4-bit) by ibm-granite",
          "build_page": "/inferences/granite-4-2/3b/ibm-granite-q4_k_m-ollama",
          "ref": "hf.co/ibm-granite/granite-4.2-3b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 16.87,
          "cost_usd_per_million": 12.8887,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "GTgUE+M0YNBrhCd+nfhxvfuUTHXQltxaRxxqr+BRb3X6swidzvpDLWqIJLj49oUitPT6RumT6tXbUpxWniVtAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "g8K0PFW88Jhe5Bfg9xesFajjxiyXUWNxUwNdxY7RDHc="
          },
          "tokens_per_usd_hour": 22
        },
        {
          "model": "Ministral 3 3B-Instruct-2512",
          "page": "/inferences/ministral-3/3b-instruct-2512",
          "build": "Q4_K_M (4-bit) by mistralai",
          "build_page": "/inferences/ministral-3/3b-instruct-2512/mistralai-q4_k_m-ollama",
          "ref": "hf.co/mistralai/Ministral-3-3B-Instruct-2512-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 16.51,
          "cost_usd_per_million": 13.1616,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 65536,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "M9RX6WM7k/iuAUJBwS3JEJChcUEJjDx8CQpa9aw+p39b9GuIjQXlqHLxpVf8AzICNErJY0ImEFmR7i4L8XOQBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "VivKeR6liU0dRGiOvj7R5OPjDCMZqoxAQEwngG/j88w="
          },
          "tokens_per_usd_hour": 21
        },
        {
          "model": "Krutrim-2 Instruct",
          "page": "/inferences/krutrim/2",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/krutrim/2/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/krutrim-ai-labs_Krutrim-2-instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 14.49,
          "cost_usd_per_million": 15.2823,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "z44e0c3eU9BSOGCqW9ua6PRQDVs1iqeYn9wq7ahuJ9mQfIlu92QHj59qz16RACFmfZ7K/Lm4PdHce4DJzwEFBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "tvRXXqsyBiV/C6V0yFMqPnUmgvRxaGwp8JqF6Ipz86w="
          },
          "tokens_per_usd_hour": 18
        },
        {
          "model": "MiniCPM-V 4.5 Base",
          "page": "/inferences/minicpm-v-4-5/base",
          "build": "Q4_K_M (4-bit) by openbmb",
          "build_page": "/inferences/minicpm-v-4-5/base/openbmb-q4_k_m-ollama",
          "ref": "hf.co/openbmb/MiniCPM-V-4_5-gguf:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 11.32,
          "cost_usd_per_million": 19.403,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 16384,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "o3wXUzdllDEGjunbws96CoEA+QpKgELnVgd1uyT4SNCcT8Hd5DBRPua9GjFy/f0t2fH7VI0reYINkoj5ootXCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "0lOoDd82ljreZYZRB/Vhquv8nK+wM2Jwuu1gPoyXKYM="
          },
          "tokens_per_usd_hour": 14
        },
        {
          "model": "Muse Glimmer 30B",
          "page": "/inferences/muse-glimmer/30b",
          "build": "Q4_K_M (4-bit) by meta-models",
          "build_page": "/inferences/muse-glimmer/30b/meta-models-q4_k_m-ollama",
          "ref": "hf.co/meta-models/Muse-Glimmer-30B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 6.9,
          "cost_usd_per_million": 33.14,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "dMDssy8soIcd+lIHE7HG9kD6lZsN72dm/zX2ExyZ59IHaBXxamFT8D/4iu4pbTICqGeh7RrvtCWHADlDPOeSDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "6ja5GGccIEW2JG1KsRCCNkZBw/mCaryhbZ7uaY17pAk="
          },
          "tokens_per_usd_hour": 8
        },
        {
          "model": "Qwen 3.6 27B",
          "page": "/inferences/qwen-3-6/27b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/qwen-3-6/27b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Qwen3.6-27B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 5.53,
          "cost_usd_per_million": 41.3775,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "KsOXom9l/c3ZoV84YzBIIfgMIRYd5X1jhFZRyY3gm7ZA788z5jOt5lOHbSayVZgTrFlHQQ23zApSBMfZPqXdBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "vGPQHnVhlttwDl31XvKYhRNHFw08oGNIOtbhSMsjYr8="
          },
          "tokens_per_usd_hour": 7
        },
        {
          "model": "Magistral Small 2506",
          "page": "/inferences/magistral-small/2506",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/magistral-small/2506/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Magistral-Small-2506-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 5.48,
          "cost_usd_per_million": 41.3999,
          "measured_at": "2026-11-07T05:09:52.970Z",
          "context_tokens": 40960,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "CoalCL8sFFUAgecKAqqZEOc8Ffy1bOMs0FsE8Z9d6BuEzAjIwotV8WoFPi1jeKMexNKqihBx+TKKr+Qci2ozAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "UpvV+HGgEH5SK4FNK981w5hpcktLBxbtSdgiIQRStF8="
          },
          "tokens_per_usd_hour": 7
        }
      ],
      "rank": 3,
      "of": 36,
      "percentiles": {
        "gpu_tflops": 60,
        "memory_bandwidth": 59,
        "cpu_all_cores": 68,
        "cpu_one_core": 61,
        "disk_read": 54,
        "download": 80
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "9e3c79",
      "did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
      "name": "Gram ·9e3c79",
      "claimed": true,
      "online": false,
      "last_seen": "2026-11-15T08:13:46.053Z",
      "version": "f3f806e",
      "region": {
        "code": "IN_562114",
        "city": "Bengaluru North",
        "state": "Karnataka"
      },
      "hardware": {
        "os": "windows",
        "arch": "x86_64",
        "class": "windows/amd64/intel/16g",
        "cpu_model": "12th Gen Intel Core i3-12100",
        "cpu_clock_ghz": null,
        "cpus": 8,
        "ram_gb": 15.747482299804688,
        "memory_bandwidth_gb_s": 4.6,
        "gpu": "Intel(R) UHD Graphics 730",
        "vram_gb": 1.9999961853027344,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 2237.9,
        "gpu_count": 1,
        "gpu_org": "intel",
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "12th Gen Intel Core i3-12100",
          "model_raw": "12th Gen Intel Core i3-12100",
          "clock_ghz": null,
          "cores_logical": 8,
          "cores_physical": null,
          "threads_per_core": null,
          "sockets": null,
          "l2_kb": null,
          "l3_mb": null,
          "single_core_hashes_per_sec": 1697.7,
          "multi_core_hashes_per_sec": 7466.7,
          "sustained_ratio": 0.9,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-15T04:07:55.704Z",
          "measured_in": "IN_562114",
          "signed": true
        },
        "memory": {
          "total_gb": 15.7,
          "bandwidth_gb_s": 4.6,
          "latency_ns": 21,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-15T04:07:56.290Z",
          "measured_in": "IN_562114",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 2237.9,
          "write_mb_s": 323.5,
          "random_read_iops": 149734,
          "fs_type": null,
          "device_model": null,
          "measured_at": "2026-11-15T04:07:56.778Z",
          "measured_in": "IN_562114",
          "signed": true
        }
      },
      "signed_runs": 67,
      "signed_runs_30d": 43,
      "builds_measured": 67,
      "best_results": 16,
      "first_signed_at": "2026-10-28T15:47:59.988Z",
      "last_signed_at": "2026-11-15T06:46:12.289Z",
      "awake_pct": 100,
      "queue": {
        "measured": 4,
        "failed": 11,
        "queued": 1531
      },
      "download_mib_per_sec": 4.3,
      "uptime_days": [
        {
          "date": "2026-10-28",
          "hours": 2
        },
        {
          "date": "2026-10-29",
          "hours": 12
        },
        {
          "date": "2026-10-30",
          "hours": 14
        },
        {
          "date": "2026-10-31",
          "hours": 6
        },
        {
          "date": "2026-11-01",
          "hours": 6
        },
        {
          "date": "2026-11-02",
          "hours": 5
        },
        {
          "date": "2026-11-03",
          "hours": 9
        },
        {
          "date": "2026-11-04",
          "hours": 11
        },
        {
          "date": "2026-11-05",
          "hours": 8
        },
        {
          "date": "2026-11-06",
          "hours": 10
        },
        {
          "date": "2026-11-07",
          "hours": 7
        },
        {
          "date": "2026-11-08",
          "hours": 13
        },
        {
          "date": "2026-11-09",
          "hours": 14
        },
        {
          "date": "2026-11-10",
          "hours": 11
        },
        {
          "date": "2026-11-11",
          "hours": 12
        },
        {
          "date": "2026-11-12",
          "hours": 17
        },
        {
          "date": "2026-11-13",
          "hours": 24
        },
        {
          "date": "2026-11-14",
          "hours": 24
        },
        {
          "date": "2026-11-15",
          "hours": 9
        }
      ],
      "uptime_pct": 46.9,
      "best_value": {
        "model": "Gemma 3 270M",
        "page": "/inferences/gemma-3/270m",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
        "engine": "ollama",
        "tokens_per_sec": 51.51,
        "cost_usd_per_million": 1.3816,
        "measured_at": "2026-11-15T06:46:12.289Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
          "signature": "zkcES8cXZ7skzNxSnPib3/T8sm2V2+s6cNUllSKND5IaWSR+SfJjOG/AmbMbCZoTWpiwkoiEzDy5W5zNbnwzAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "3Mw+WsNzZKxl/2a2mJc2kHTEXsWOP/Q4ICa38Jn9DBQ="
        },
        "tokens_per_usd_hour": 201
      },
      "opened": [
        {
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q2_K_L",
          "family": "phi-4",
          "member": "mini-reasoning",
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q4_K_M",
          "family": "phi-4",
          "member": "mini-reasoning",
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "gemma4:e2b-it-qat",
          "family": "gemma-4",
          "member": "e2b",
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "google/gemma-4-E2B-it-qat-q4_0-gguf",
          "family": "gemma-4",
          "member": "e2b",
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Qwen3-0.6B-GGUF:Q3_K_M",
          "family": "qwen-3",
          "member": "0.6b",
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Qwen3-0.6B-GGUF:Q4_K_M",
          "family": "qwen-3",
          "member": "0.6b",
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Qwen3-0.6B-GGUF:Q5_K_M",
          "family": "qwen-3",
          "member": "0.6b",
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "hf.co/Mungert/Phi-4-mini-instruct.gguf:IQ2_XS",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q3_K_S",
          "family": "gemma-4",
          "member": "e2b",
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3-VL-4B-Instruct-GGUF:IQ4_XS",
          "family": "qwen-3",
          "member": "4b",
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3-VL-4B-Instruct-GGUF:Q2_K",
          "family": "qwen-3",
          "member": "4b",
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "lfm2.5-thinking:1.2b",
          "family": null,
          "member": null,
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "phi3:3.8b",
          "family": null,
          "member": null,
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "qwen3:0.6b",
          "family": "qwen-3",
          "member": "0.6b",
          "at": "2026-11-15T06:46:12.289Z"
        },
        {
          "ref": "tinyllama:1.1b",
          "family": null,
          "member": null,
          "at": "2026-11-15T06:46:12.289Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-28",
          "runs": 80
        },
        {
          "date": "2026-10-30",
          "runs": 1200
        },
        {
          "date": "2026-11-01",
          "runs": 80
        },
        {
          "date": "2026-11-02",
          "runs": 480
        },
        {
          "date": "2026-11-03",
          "runs": 1480
        },
        {
          "date": "2026-11-04",
          "runs": 2480
        },
        {
          "date": "2026-11-05",
          "runs": 1560
        },
        {
          "date": "2026-11-06",
          "runs": 880
        },
        {
          "date": "2026-11-07",
          "runs": 2739
        },
        {
          "date": "2026-11-08",
          "runs": 1524
        },
        {
          "date": "2026-11-09",
          "runs": 2494
        },
        {
          "date": "2026-11-10",
          "runs": 1215
        },
        {
          "date": "2026-11-11",
          "runs": 2683
        },
        {
          "date": "2026-11-12",
          "runs": 1450
        },
        {
          "date": "2026-11-13",
          "runs": 1913
        },
        {
          "date": "2026-11-14",
          "runs": 2179
        },
        {
          "date": "2026-11-15",
          "runs": 90
        }
      ],
      "runs_by_family": [
        {
          "family": "qwen-3",
          "runs": 11
        },
        {
          "family": "gemma-4",
          "runs": 9
        },
        {
          "family": "llama-3-2",
          "runs": 7
        },
        {
          "family": "other",
          "runs": 7
        },
        {
          "family": "phi-4",
          "runs": 5
        },
        {
          "family": "depth-anything",
          "runs": 2
        },
        {
          "family": "embeddinggemma",
          "runs": 2
        },
        {
          "family": "cogvideox",
          "runs": 2
        },
        {
          "family": "whisper",
          "runs": 2
        },
        {
          "family": "stable-audio-open",
          "runs": 2
        },
        {
          "family": "musicgen",
          "runs": 1
        },
        {
          "family": "sam",
          "runs": 1
        },
        {
          "family": "qwen-2-5",
          "runs": 1
        },
        {
          "family": "nllb",
          "runs": 1
        },
        {
          "family": "sarvam",
          "runs": 1
        },
        {
          "family": "tiny-sd",
          "runs": 1
        },
        {
          "family": "sd-turbo",
          "runs": 1
        },
        {
          "family": "gemma-3",
          "runs": 1
        },
        {
          "family": "kokoro",
          "runs": 1
        },
        {
          "family": "moonshine",
          "runs": 1
        },
        {
          "family": "f5-tts",
          "runs": 1
        },
        {
          "family": "sana",
          "runs": 1
        },
        {
          "family": "minicpm-5",
          "runs": 1
        },
        {
          "family": "hy-mt-2",
          "runs": 1
        },
        {
          "family": "minicpm-v-4-6",
          "runs": 1
        },
        {
          "family": "wan-2",
          "runs": 1
        },
        {
          "family": "qwen-image",
          "runs": 1
        },
        {
          "family": "ltx-video",
          "runs": 1
        }
      ],
      "rooms": [
        "26NRJ3lXsSmqiI4k6N4X",
        "IN_752115"
      ],
      "awards": [
        {
          "id": "runs-10",
          "label": "10+ signed runs"
        },
        {
          "id": "opener",
          "label": "First to measure 15 builds"
        },
        {
          "id": "best-10",
          "label": "Best result on 16 builds"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "Gemma 3 270M",
        "page": "/inferences/gemma-3/270m",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
        "engine": "ollama",
        "tokens_per_sec": 51.51,
        "cost_usd_per_million": 1.3816,
        "measured_at": "2026-11-15T06:46:12.289Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
          "signature": "zkcES8cXZ7skzNxSnPib3/T8sm2V2+s6cNUllSKND5IaWSR+SfJjOG/AmbMbCZoTWpiwkoiEzDy5W5zNbnwzAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "3Mw+WsNzZKxl/2a2mJc2kHTEXsWOP/Q4ICa38Jn9DBQ="
        },
        "tokens_per_usd_hour": 201
      },
      "cheapest": {
        "model": "Gemma 3 270M",
        "page": "/inferences/gemma-3/270m",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
        "engine": "ollama",
        "tokens_per_sec": 51.51,
        "cost_usd_per_million": 1.3816,
        "measured_at": "2026-11-15T06:46:12.289Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
          "signature": "zkcES8cXZ7skzNxSnPib3/T8sm2V2+s6cNUllSKND5IaWSR+SfJjOG/AmbMbCZoTWpiwkoiEzDy5W5zNbnwzAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "3Mw+WsNzZKxl/2a2mJc2kHTEXsWOP/Q4ICa38Jn9DBQ="
        },
        "tokens_per_usd_hour": 201
      },
      "builds": [
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 51.51,
          "cost_usd_per_million": 1.3816,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "zkcES8cXZ7skzNxSnPib3/T8sm2V2+s6cNUllSKND5IaWSR+SfJjOG/AmbMbCZoTWpiwkoiEzDy5W5zNbnwzAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "3Mw+WsNzZKxl/2a2mJc2kHTEXsWOP/Q4ICa38Jn9DBQ="
          },
          "tokens_per_usd_hour": 201
        },
        {
          "model": "Qwen2.5 0.5B Instruct",
          "page": "/inferences/qwen-2-5/0.5b",
          "build": "Q4_K_M (4-bit) by Qwen",
          "build_page": "/inferences/qwen-2-5/0.5b/qwen-q4_k_m-ollama-148138f0",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 25.57,
          "cost_usd_per_million": 2.791,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "a8iCpfgwDpzHf+Mqc4d8mPd4RS4tj8MYRTmwI25HVFD0EdHCGjLAWDGRARoC5kJ7qhP0FCr3rDcyLeW7K2sFDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "rynD1JdS2lmJ50D8LUxRMmTYnUV7BCgc7PhDx7IDmeM="
          },
          "tokens_per_usd_hour": 100
        },
        {
          "model": "MiniCPM5 1B",
          "page": "/inferences/minicpm-5/1b",
          "build": "Q4_K_M (4-bit) by openbmb",
          "build_page": "/inferences/minicpm-5/1b/openbmb-q4_k_m-ollama",
          "ref": "hf.co/openbmb/MiniCPM5-1B-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 21.73,
          "cost_usd_per_million": 3.2909,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "axoj7LjWYgugssNO1lXIf75xcEnquLUZgHdQwZiziQpUeJuDvYzpe3+Op+Uy4fc0YzRKdyxWQ+ICVaD6V22hAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "roAsaoDsgMIV2mA3qhUMZCSnR6/rHTR5v2CdwwEMs5U="
          },
          "tokens_per_usd_hour": 84
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": "IQ3_XS (3-bit) by maziyarpanahi",
          "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-iq3_xs-ollama",
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:IQ3_XS",
          "engine": "ollama",
          "tokens_per_sec": 17.72,
          "cost_usd_per_million": 4.0054,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "WqKWZGWm3xXMnhC4j4wDAQTBg1V3Xz930DW5aKw9lPqBOK0RGyn8kqgRg/OAMw+zHviu9JeGTlmk3/QlaIgvDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "HHahuvMAN84Rid0xaed1SEw5x/umOUQUKwUaH/9YE3E="
          },
          "tokens_per_usd_hour": 69
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q3_K_S (3-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q3_k_s-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q3_K_S",
          "engine": "ollama",
          "tokens_per_sec": 13,
          "cost_usd_per_million": 5.5982,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "ciAHBLyGB79ij368pSfUFAczmAnHBePLvRuSY/iXnGm7RZFvDexxcDWgQ1DMTP84U2OLlgoXQK6suaIEWN/eCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "HEMTO6qC/psFOEv69Hs+ibyhavMscQB5YyFsCNa1bbw="
          },
          "tokens_per_usd_hour": 50
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/llama-3-2/1b/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 11.62,
          "cost_usd_per_million": 6.161,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "wDTHJzoIikB2dE6nDNndfzo2VtP9Jhv7K3dpiJmPAnaypEJrW8jge5+iLBanTuViujbRJxKWT1o3DkQDLtHsCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "uUiXSylyDMFzJvJCGvFj4j9UR78WmdJTQJwGMDqTtC0="
          },
          "tokens_per_usd_hour": 45
        },
        {
          "model": "Hunyuan MT 2 1.8B",
          "page": "/inferences/hy-mt-2/1.8b",
          "build": "Q4_K_M (4-bit) by tencent",
          "build_page": "/inferences/hy-mt-2/1.8b/tencent-q4_k_m-ollama",
          "ref": "hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 10.63,
          "cost_usd_per_million": 6.7573,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "1I3J15zrqtK2sw2NQWhjpvF13Q1dcC0YPfCFK2FCEdDTvcB0XNrgKgz4DNW/w7qv7tEevciL6xYWVJO2mxFRBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "38UlxBuYq+4zsiivvyPkH/7oM2pcA/XDsdw6zV3KGRM="
          },
          "tokens_per_usd_hour": 41
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "Q2_K (2-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-q2_k-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q2_K",
          "engine": "ollama",
          "tokens_per_sec": 9.83,
          "cost_usd_per_million": 7.3078,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "b/w+k4PL1xv5wjvHmhr8aXL+R/9fN/y4D7vS4+w0plH5EcOXRSkVPQ+X1uPkAixob3TQaRKgUOCLcwNgNpwcDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "CeE2caDeuSRVRHmWnmkta8KGkKBOt9jqSe6lkM4Skvo="
          },
          "tokens_per_usd_hour": 38
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_0 (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_0-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_0",
          "engine": "ollama",
          "tokens_per_sec": 9.1,
          "cost_usd_per_million": 7.9949,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "C1x1pWyS4Hl82iIs75U95aAu5WSkwkVDcdmJr7T0FfNrm8lXjp/7vjh4IyIH1n0xpFcDdvuiqVt18a2N0p+vBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "yZ/NXHXpTosnejliBa+6+KvbkNJv7jc7lMPCzd2Qpyo="
          },
          "tokens_per_usd_hour": 35
        },
        {
          "model": "Sarvam-1",
          "page": "/inferences/sarvam/1",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 8.1,
          "cost_usd_per_million": 8.9138,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "4AF9srIpWtrtXuqApmRSZDoZwf5KezUUcayXypSU4T5NTfUInkHA6t9baB2CQh1wAXvq/4LgCm+oG8S+QUCSBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "8Y5eP0X7r/hzHXpQuDYRvLz+ATLb03Ok2l3VDq0BRdQ="
          },
          "tokens_per_usd_hour": 31
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama",
          "tokens_per_sec": 6.8,
          "cost_usd_per_million": 10.9087,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "BMywkb7dV7N0NCPCAwy2frRvD+1b5KnYKmhCfTLOR3UmayTD4dvFcTYzmq6G85h+OCYe1P+rzHd/GFQsWLZOBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "2BWBi/8yZXNYa1whXNPsNiCfdhfLDrd6wTyXZiJQjI8="
          },
          "tokens_per_usd_hour": 25
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 6.5,
          "cost_usd_per_million": 11.221,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "WZAvN6ZkpZQC4GFjGuWfJfOE4iB7/kpD/hvNHu1bPjwOd6U3N4NAwNMbgHrW9mMBNQJ/vjHlP7JzWYR7OrdqAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "W/IEWu7EX7sLeyrVp1DbaZYSgJUfhLyVtlcPjRTahoo="
          },
          "tokens_per_usd_hour": 25
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "IQ1_S (1-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq1_s-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ1_S",
          "engine": "ollama",
          "tokens_per_sec": 5.6,
          "cost_usd_per_million": 12.8228,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "8fcnRaQ3m3MA0c8FUGdMkij/UPytVibwj2cQdk1FKhBz7CDPbcnWm2se6wyNxw+nyoYYQO1NYS++pg7w8TYLCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "vmXoshOOe7Fs0NKATmXey2PaLZee+Nw0RYgUwyEVb+Y="
          },
          "tokens_per_usd_hour": 22
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/llama-3-2/3b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 5.03,
          "cost_usd_per_million": 14.4133,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "GL9Os9TZx4O7LCb/cOIjuGxtDw4mfdMIdVpPNwwMJ0/VDkm4WdOC9wesPm8EbIVC3y648rsZAthcVUEzUJJRBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "si2oJrhjr04zYCCKf9m16p0R77k8ZYclC4KtWepKxqg="
          },
          "tokens_per_usd_hour": 19
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "IQ3_XXS (3-bit) by unsloth",
          "build_page": "/inferences/qwen-3/4b/unsloth-iq3_xxs-ollama",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ3_XXS",
          "engine": "ollama",
          "tokens_per_sec": 4.63,
          "cost_usd_per_million": 15.5931,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "Y59PBxXEnDHJacJxLxT+HCiEDoe34nJEOv9xMt6Y6hZV36f35FyE5ZbfZvbL3vdcW91Mq9p8u4xqxvlHJlmeAA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NC42MzQ4MTkxNDg5NTI1MTcsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0NzI1MTcyMjg5MzMyMDAwLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwM2ZlNjljZTQ0ZDdiMTQxOGRkNDI2MzMxYTUzOWYxMmMxZGU0MjhhMjM1MDE3NDQyZDQxY2VhYjBhMjllM2M3OSIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoid2luZG93cy9hbWQ2NC9pbnRlbC8xNmciLCJtb2RlbCI6ImhmLmNvL3Vuc2xvdGgvUXdlbjMtNEItR0dVRjpJUTNfWFhTIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hIiwic2FtcGxlX2NvdW50IjoyODAsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6MjI1ODYuNDI2MDU3OTk2MDA2LCJ0b3RhbF90b2tlbnMiOjEwNDY4NH19",
            "payload_sha256": "WbrJ97rZca2bL/pmCLZbmIP7nFXQ8I9bar4brJJnAPA="
          },
          "tokens_per_usd_hour": 18
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "IQ2_XS (2-bit) by mungert",
          "build_page": "/inferences/phi-4/mini/mungert-iq2_xs-ollama",
          "ref": "hf.co/Mungert/Phi-4-mini-instruct.gguf:IQ2_XS",
          "engine": "ollama",
          "tokens_per_sec": 4.41,
          "cost_usd_per_million": 16.1008,
          "measured_at": "2026-11-15T06:46:12.289Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "oRPz1Ag39CANoFCV2iueHhXS+bHpN8bRBOpUlfiru6y9bY3EjNjPRX18+BkIVxPEfqKJ0Jvu+b333nJMwvFmBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "NwtWyN5ntWcpOP1iFxECpVTLIoSlYWtmnGeFS74tAlg="
          },
          "tokens_per_usd_hour": 17
        }
      ],
      "rank": 4,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 70,
        "cpu_all_cores": 82,
        "cpu_one_core": 86,
        "disk_read": 43,
        "download": 10
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "84bf54",
      "did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
      "name": "Gram ·84bf54",
      "claimed": true,
      "online": false,
      "last_seen": "2026-11-10T11:57:23.264Z",
      "version": "dbf9ea5",
      "region": {
        "code": "IN_751030",
        "city": "Bhubaneswar",
        "state": "Odisha"
      },
      "hardware": {
        "os": "windows",
        "arch": "x86_64",
        "class": "windows/amd64/nvidia/16g",
        "cpu_model": "AMD Ryzen 7 5800X 8-Core",
        "cpu_clock_ghz": null,
        "cpus": 16,
        "ram_gb": 15.906551361083984,
        "memory_bandwidth_gb_s": 15.3,
        "gpu": "NVIDIA GeForce RTX 3080 Ti",
        "vram_gb": 12,
        "gpu_tflops_fp16": 59.8,
        "disk_read_mb_s": 801.2,
        "gpu_count": 1,
        "gpu_org": "nvidia",
        "cpu_org": "amd"
      },
      "sheet": {
        "cpu": {
          "model": "AMD Ryzen 7 5800X 8-Core",
          "model_raw": "AMD Ryzen 7 5800X 8-Core",
          "clock_ghz": null,
          "cores_logical": 16,
          "cores_physical": null,
          "threads_per_core": null,
          "sockets": null,
          "l2_kb": null,
          "l3_mb": null,
          "single_core_hashes_per_sec": 2085.3,
          "multi_core_hashes_per_sec": 26483.6,
          "sustained_ratio": 1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-10T09:44:03.726Z",
          "measured_in": "IN_751030",
          "signed": true
        },
        "memory": {
          "total_gb": 15.9,
          "bandwidth_gb_s": 15.3,
          "latency_ns": 11.8,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-10T09:44:04.282Z",
          "measured_in": "IN_751030",
          "signed": true
        },
        "gpu": {
          "name": "NVIDIA GeForce RTX 3080 Ti",
          "memory_gb": 12,
          "fp16_tflops": 59.8,
          "watts": null,
          "tflops_per_watt": null,
          "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
          "measured_at": "2026-11-10T09:44:16.445Z",
          "measured_in": "IN_751030",
          "signed": true
        },
        "storage": {
          "read_mb_s": 801.2,
          "write_mb_s": 73.5,
          "random_read_iops": 4296,
          "fs_type": null,
          "device_model": null,
          "measured_at": "2026-11-10T09:44:05.469Z",
          "measured_in": "IN_751030",
          "signed": true
        }
      },
      "signed_runs": 51,
      "signed_runs_30d": 34,
      "builds_measured": 51,
      "best_results": 33,
      "first_signed_at": "2026-10-28T05:58:39.086Z",
      "last_signed_at": "2026-11-10T11:20:52.995Z",
      "awake_pct": 100,
      "queue": {
        "measured": 10,
        "failed": 13,
        "queued": 1392
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-11-08",
          "hours": 11
        },
        {
          "date": "2026-11-10",
          "hours": 4
        }
      ],
      "uptime_pct": 31.3,
      "best_value": {
        "model": "DeepSeek-OCR 2 Base",
        "page": "/inferences/deepseek-ocr-2/base",
        "build": "Q4_K_M (4-bit) by sabafallah",
        "build_page": "/inferences/deepseek-ocr-2/base/sabafallah-q4_k_m-ollama",
        "ref": "hf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_M",
        "engine": "ollama",
        "tokens_per_sec": 411.73,
        "cost_usd_per_million": 0.626,
        "measured_at": "2026-11-10T11:20:52.995Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "4HYRG5+mdqESuBQ4ErLCPQPGiOwQDxPVtoxq6zDaTvsKyh7lrdfn21E4ptnYfxsHf9tne9OYhAsensMcYCbmBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "77j02NRghpxfYrHhfhcXztyhdwWND1BnRDyCcC+fHdo="
        },
        "tokens_per_usd_hour": 444
      },
      "opened": [
        {
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:BF16",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-10T11:20:52.995Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "family": null,
          "member": null,
          "at": "2026-11-10T11:20:52.995Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q2_K",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-10T11:20:52.995Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q4_K_M",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-10T11:20:52.995Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q4_K_S",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-10T11:20:52.995Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q8_0",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-10T11:20:52.995Z"
        },
        {
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:F16",
          "family": "llama-3-2",
          "member": "3b",
          "at": "2026-11-10T11:20:52.995Z"
        },
        {
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ2_XXS",
          "family": "llama-3-2",
          "member": "3b",
          "at": "2026-11-10T11:20:52.995Z"
        },
        {
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ3_XXS",
          "family": "llama-3-2",
          "member": "3b",
          "at": "2026-11-10T11:20:52.995Z"
        },
        {
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_0",
          "family": "llama-3-2",
          "member": "3b",
          "at": "2026-11-10T11:20:52.995Z"
        },
        {
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q8_0",
          "family": "llama-3-2",
          "member": "3b",
          "at": "2026-11-10T11:20:52.995Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:BF16",
          "family": "qwen-3",
          "member": "4b",
          "at": "2026-11-10T11:20:52.995Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-28",
          "runs": 29
        },
        {
          "date": "2026-11-08",
          "runs": 291
        },
        {
          "date": "2026-11-09",
          "runs": 34
        },
        {
          "date": "2026-11-10",
          "runs": 209
        }
      ],
      "runs_by_family": [
        {
          "family": "phi-4",
          "runs": 10
        },
        {
          "family": "qwen-3",
          "runs": 8
        },
        {
          "family": "llama-3-2",
          "runs": 8
        },
        {
          "family": "gemma-4",
          "runs": 3
        },
        {
          "family": "depth-anything",
          "runs": 2
        },
        {
          "family": "nllb",
          "runs": 2
        },
        {
          "family": "musicgen",
          "runs": 1
        },
        {
          "family": "sam",
          "runs": 1
        },
        {
          "family": "qwen-2-5",
          "runs": 1
        },
        {
          "family": "sarvam",
          "runs": 1
        },
        {
          "family": "tiny-sd",
          "runs": 1
        },
        {
          "family": "sd-turbo",
          "runs": 1
        },
        {
          "family": "gemma-3",
          "runs": 1
        },
        {
          "family": "mms-tts",
          "runs": 1
        },
        {
          "family": "sana",
          "runs": 1
        },
        {
          "family": "minicpm-5",
          "runs": 1
        },
        {
          "family": "embeddinggemma",
          "runs": 1
        },
        {
          "family": "hy-mt-2",
          "runs": 1
        },
        {
          "family": "cogvideox",
          "runs": 1
        },
        {
          "family": "seamless-m4t",
          "runs": 1
        },
        {
          "family": "deepseek-ocr-2",
          "runs": 1
        },
        {
          "family": "wan-2",
          "runs": 1
        },
        {
          "family": "qwen-image",
          "runs": 1
        },
        {
          "family": "other",
          "runs": 1
        }
      ],
      "rooms": [
        "DG1h70GzVONqlxjluvnM",
        "IN_110054",
        "IN_751001",
        "IN_751002",
        "IN_751030",
        "IN_752050",
        "IN_752101",
        "j9p5QqSCECG55k3Qf98Z"
      ],
      "awards": [
        {
          "id": "runs-10",
          "label": "10+ signed runs"
        },
        {
          "id": "opener",
          "label": "First to measure 12 builds"
        },
        {
          "id": "best-10",
          "label": "Best result on 33 builds"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "DeepSeek-OCR 2 Base",
        "page": "/inferences/deepseek-ocr-2/base",
        "build": "Q4_K_M (4-bit) by sabafallah",
        "build_page": "/inferences/deepseek-ocr-2/base/sabafallah-q4_k_m-ollama",
        "ref": "hf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_M",
        "engine": "ollama",
        "tokens_per_sec": 411.73,
        "cost_usd_per_million": 0.626,
        "measured_at": "2026-11-10T11:20:52.995Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "4HYRG5+mdqESuBQ4ErLCPQPGiOwQDxPVtoxq6zDaTvsKyh7lrdfn21E4ptnYfxsHf9tne9OYhAsensMcYCbmBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "77j02NRghpxfYrHhfhcXztyhdwWND1BnRDyCcC+fHdo="
        },
        "tokens_per_usd_hour": 444
      },
      "cheapest": {
        "model": "DeepSeek-OCR 2 Base",
        "page": "/inferences/deepseek-ocr-2/base",
        "build": "Q4_K_M (4-bit) by sabafallah",
        "build_page": "/inferences/deepseek-ocr-2/base/sabafallah-q4_k_m-ollama",
        "ref": "hf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_M",
        "engine": "ollama",
        "tokens_per_sec": 411.73,
        "cost_usd_per_million": 0.626,
        "measured_at": "2026-11-10T11:20:52.995Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "4HYRG5+mdqESuBQ4ErLCPQPGiOwQDxPVtoxq6zDaTvsKyh7lrdfn21E4ptnYfxsHf9tne9OYhAsensMcYCbmBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "77j02NRghpxfYrHhfhcXztyhdwWND1BnRDyCcC+fHdo="
        },
        "tokens_per_usd_hour": 444
      },
      "builds": [
        {
          "model": "DeepSeek-OCR 2 Base",
          "page": "/inferences/deepseek-ocr-2/base",
          "build": "Q4_K_M (4-bit) by sabafallah",
          "build_page": "/inferences/deepseek-ocr-2/base/sabafallah-q4_k_m-ollama",
          "ref": "hf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 411.73,
          "cost_usd_per_million": 0.626,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "4HYRG5+mdqESuBQ4ErLCPQPGiOwQDxPVtoxq6zDaTvsKyh7lrdfn21E4ptnYfxsHf9tne9OYhAsensMcYCbmBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "77j02NRghpxfYrHhfhcXztyhdwWND1BnRDyCcC+fHdo="
          },
          "tokens_per_usd_hour": 444
        },
        {
          "model": "MiniCPM5 1B",
          "page": "/inferences/minicpm-5/1b",
          "build": "Q4_K_M (4-bit) by openbmb",
          "build_page": "/inferences/minicpm-5/1b/openbmb-q4_k_m-ollama",
          "ref": "hf.co/openbmb/MiniCPM5-1B-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 381.36,
          "cost_usd_per_million": 0.6733,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "ozh3hx0/d7WPcYqg+IkhH50y8z/CTeLYGY4A2j49FKW0nGMHRvC/A+sw3CvrGSTdkiErOw1Lqp9qoKP3Vw6KDA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MzgxLjM2MTM5MTE2NDYxNzMsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0MzA5NjUyOTk1MDMyMTAwLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwYmM2NTBiNTlhZjg3OTQ4OTYwMzFkMmU4MmJjYTBlMTE5ZjM5MGUyMWQ2OTA5NjNhZTQ3NTc4MjM4MDg0YmY1NCIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoid2luZG93cy9hbWQ2NC9udmlkaWEvMTZnIiwibW9kZWwiOiJoZi5jby9vcGVuYm1iL01pbmlDUE01LTFCLUdHVUY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hIiwic2FtcGxlX2NvdW50IjozNiwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjo0Ny41MzQ5NjM5OTk5OTk5OTUsInRvdGFsX3Rva2VucyI6MTgxMjh9fQ==",
            "payload_sha256": "LnhDHC/Su98gwUes4AKE6RB5M1olge99nDgeKxvyc1o="
          },
          "tokens_per_usd_hour": 413
        },
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 286.08,
          "cost_usd_per_million": 0.8964,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "gHIs9srClOIWui6TB0a8u3+HYoQ2InpMQsoIQ1ZePCmEE6rV/FV0eksjOcgLUbFGt48dsyxNJ87TFrKN7/PEAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "Mo4jh+tDGOYKA4gL1wH8vXyFhpK+oQvtLnWGcnukMQs="
          },
          "tokens_per_usd_hour": 310
        },
        {
          "model": "Qwen2.5 0.5B Instruct",
          "page": "/inferences/qwen-2-5/0.5b",
          "build": "Q4_K_M (4-bit) by Qwen",
          "build_page": "/inferences/qwen-2-5/0.5b/qwen-q4_k_m-ollama-148138f0",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 247.91,
          "cost_usd_per_million": 1.0352,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "zHgYIgtg5QgGhL3uf7V0VLRa/7/tlXLVJEZanHeu2Gwz89AeOtWygupmS8NxZVNrrm8hZsz70TkU4FIWXe9+Cg==",
            "signing_payload_b64": "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",
            "payload_sha256": "tNTIseiRt0C5b9dIx68HC99rXoydMVQFXz7SN7cITcE="
          },
          "tokens_per_usd_hour": 268
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "IQ1_S (1-bit) by unsloth",
          "build_page": "/inferences/llama-3-2/3b/unsloth-iq1_s-ollama",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ1_S",
          "engine": "ollama",
          "tokens_per_sec": 225.19,
          "cost_usd_per_million": 1.1411,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "PveBSNBhE0lObW4aP806WVjqea6j8d4LxYTNYJmQWaKaIPSAns5BY2lQrmwh0hVRGSUW1mbi9QGmp9L1kdYoDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "79WvM1JAceTQ8k+h2yLBG41bLAwQYkVveC5r27WpGzM="
          },
          "tokens_per_usd_hour": 243
        },
        {
          "model": "Hunyuan MT 2 1.8B",
          "page": "/inferences/hy-mt-2/1.8b",
          "build": "Q4_K_M (4-bit) by tencent",
          "build_page": "/inferences/hy-mt-2/1.8b/tencent-q4_k_m-ollama",
          "ref": "hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 221.76,
          "cost_usd_per_million": 1.1595,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "S5f/XRcwG7fTN1OlHLHiODerHjAf+JqE7SvC+MJE8yYYEv/ySr4ui5T6Zx7ucagCPSuAeTsojhpMsDJc+S9+CQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "pCzj+fIjZyfPuWAP9hghUgs88J6II22ffnsq0T8FTew="
          },
          "tokens_per_usd_hour": 240
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q4_K_M (4-bit) by maziyarpanahi",
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q4_k_m-ollama",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 205.49,
          "cost_usd_per_million": 1.247,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "TYEeuh3rrSdgvu/LsNQESePjsV7jbCoYVOsqpoJtatkcuOC1tx2UNb/U6u3bXevDGiqKWxEMMkYOUgpX9JI2DQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "FhWLGFN+H10MATforjUzWFeZUrrjRYW/xNW/m83asiQ="
          },
          "tokens_per_usd_hour": 223
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "Q4_0 (4-bit) by unsloth",
          "build_page": "/inferences/llama-3-2/3b/unsloth-q4_0-ollama",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_0",
          "engine": "ollama",
          "tokens_per_sec": 205.06,
          "cost_usd_per_million": 1.2569,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "N6VJZ0GRJGfittPj1S+QlKfDca2DsANkEqbsFu5oiR1QDyUs6azVaTBr90Kku1XEjCwY4CnkIK5quG1IEpLoBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "2vdc8pjZzyGvjZS3dZwlxnchlSNWNEIXaOhBMHlLIGM="
          },
          "tokens_per_usd_hour": 221
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/llama-3-2/3b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 203.94,
          "cost_usd_per_million": 1.2642,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "t8ixuFktyclVNDUZo4dKyZBjnZGFFh+UoLFgCcDoy3OQXL6wfZ/gCPvIDFFKKaR1d4irKOdUHYTcRBHtTBNTBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "q31wWJLyb4Pqc9fAEUzKmkn0Kro2XXMp8qspXtnZL4I="
          },
          "tokens_per_usd_hour": 220
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "IQ3_XXS (3-bit) by unsloth",
          "build_page": "/inferences/llama-3-2/3b/unsloth-iq3_xxs-ollama",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ3_XXS",
          "engine": "ollama",
          "tokens_per_sec": 199.51,
          "cost_usd_per_million": 1.2897,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "trRpa25ZFobEaCmnWiOPXmY4losN8RnC5CYtyRivoa22T+qICZHrHATti3MLoR+Ii1gTliDDINjwsSjgWguqAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "ueKoMj7B5IY4inlr3nWTBw0zc3UePxtpUHareQKKDvQ="
          },
          "tokens_per_usd_hour": 215
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "IQ3_XXS (3-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq3_xxs-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
          "engine": "ollama",
          "tokens_per_sec": 189.43,
          "cost_usd_per_million": 1.3595,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "kd3dBY+d0Z9SiYQZuI1Vs4Zn7iaYvzy+zc/iMfSpNu2J0mmXuOXhNqTLyks0ZNqD2jb1WmXj1IcSHOM5g7JhDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "hiu1HBlqSAgtbPIV5+zmJPT/zyjyiVdlXE6V77H83B0="
          },
          "tokens_per_usd_hour": 204
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "IQ4_XS (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-iq4_xs-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_XS",
          "engine": "ollama",
          "tokens_per_sec": 182.88,
          "cost_usd_per_million": 1.4139,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "RsptUcs06S28m0uKeWTlZS9QCPWg2PQHX0AsXewXagFWGj7jdYaQVULZ2Pf5WzK+wbsPNxE0LVc/oErokeeuAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "p8d6ff7bVFI3ZqIZCAqBkR+SgYhM4f1wn/n7zRTgckI="
          },
          "tokens_per_usd_hour": 196
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by google",
          "build_page": "/inferences/gemma-4/e2b/google-q4_k_m-ollama",
          "ref": "gemma4:e2b",
          "engine": "ollama",
          "tokens_per_sec": 170.45,
          "cost_usd_per_million": 1.5375,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "0jaTRjp3ghIwgAyLI6t5t/fAAM2aQDoaoB1Ea1xTGuE0ut1V/SB1fw/cqn99aTmDwN5aC77NLkARiTs7CQVFDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "j33xtQnznek8EvD5kEKzrJSGKY4HtWocgMM15i3H5TM="
          },
          "tokens_per_usd_hour": 181
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q2_K (2-bit) by maziyarpanahi",
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q2_k-ollama",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q2_K",
          "engine": "ollama",
          "tokens_per_sec": 154.35,
          "cost_usd_per_million": 1.6602,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "KOyxMTtxouNhSfmC7bDu0EUmNL9GQ9iazowJLfugrEgYIJy3Ql2y9i2QAlrnhnzGReyg71yTThQnhqk1KA83Bg==",
            "signing_payload_b64": "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",
            "payload_sha256": "b+ywwgvAtpfwIzpnQPj2j8BHBHtcFE/Z7RHHDrs2OLE="
          },
          "tokens_per_usd_hour": 167
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q3_K_S (3-bit) by maziyarpanahi",
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q3_k_s-ollama",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_S",
          "engine": "ollama",
          "tokens_per_sec": 134.8,
          "cost_usd_per_million": 1.901,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "cYpdYJUDoS8tfO2l7N0LI9CEPLFy5oGszNDtIcbbqwcz1z/h2jLAyPH2u9r2aXley/nzSrkZbu8/bi5vNVVaDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "dBN74GKPQzjce4PDsuzeLXznH7ImC9MNAFVOlSIw30k="
          },
          "tokens_per_usd_hour": 146
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "IQ2_XXS (2-bit) by unsloth",
          "build_page": "/inferences/llama-3-2/3b/unsloth-iq2_xxs-ollama",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ2_XXS",
          "engine": "ollama",
          "tokens_per_sec": 124.17,
          "cost_usd_per_million": 2.0703,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "D1ig34cA+0j2p63F5Jbd5y39GCd2BIK4p4MB4BmuSPcdbCUzQYPToHwaQiNF9NOP0ogzadbVJlkuH7p6CWxpCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "bX24ghwlBmpnfjqNC+tb2aPXZYqPWxMItJ2K7jkRRiE="
          },
          "tokens_per_usd_hour": 134
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q3_K_L (3-bit) by maziyarpanahi",
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q3_k_l-ollama",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_L",
          "engine": "ollama",
          "tokens_per_sec": 110.09,
          "cost_usd_per_million": 2.3277,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "7ylTDBueNvVUiJZL2O8MUujv+SpZp4bhF1QWaSIVqSjw0kP6Gk/xqeLuwNnmPF3EWL1F13cgHMnBdOn2FZPQAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "CHRwQb/D4ofELxYWOxKBsE1kJyJThAgerSzxGGZEPjA="
          },
          "tokens_per_usd_hour": 119
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/llama-3-2/3b/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 99.3,
          "cost_usd_per_million": 2.5805,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 65536,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "EGkn4eE+egBD8Zqg5buqWFG8Tq93+IMRRvCyo9MiBbVqt/RW22TI3Do7s93q/2TeilF35Mf0fW8qlHg/K4CQDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "ZJ1pHvVEwj/sL1YYDZXINiwf+RBVhKfITOZC7/TBTlw="
          },
          "tokens_per_usd_hour": 108
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "F16 (16-bit, full precision) by unsloth",
          "build_page": "/inferences/llama-3-2/3b/unsloth-f16-ollama",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:F16",
          "engine": "ollama",
          "tokens_per_sec": 94.89,
          "cost_usd_per_million": 2.7532,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 16384,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "q+/oK2hh3prtgVAG8mxMI7wphBElRTpMWoKKyd+XlDDnRPW7wh//67maA9Ym/1bB9bnZyrNh+eFqhXp0zzA8AQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "i+YMgejgL3yOJlpD1XIk2TnJkGAF+WrUX9RwaU43YSQ="
          },
          "tokens_per_usd_hour": 101
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "IQ1_S (1-bit) by unsloth",
          "build_page": "/inferences/qwen-3/4b/unsloth-iq1_s-ollama",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ1_S",
          "engine": "ollama",
          "tokens_per_sec": 90.41,
          "cost_usd_per_million": 2.8436,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "L7lCCEm+MI9Hu0jbJwY1iLvp9W/yHbc+9kH5QLIT9e1R8EUVc1C8ChUoEW661NrzTEZVto8F2PDvmXS+XfZjDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "kM8rfi2mi4PJ1JVRGUvALBPl7iHN7E1HH64q5j9ZKIQ="
          },
          "tokens_per_usd_hour": 98
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "Q3_K_L (3-bit) by lmstudio-community",
          "build_page": "/inferences/phi-4/mini-reasoning/lmstudio-community-q3_k_l-ollama",
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q3_K_L",
          "engine": "ollama",
          "tokens_per_sec": 89.92,
          "cost_usd_per_million": 2.8497,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "MNjOUJ1l6SEBe4q5ADBGa1k3Mes7L+Mc4zpyIXCEtRp8wPvYrGqc0LaTh/ep7kJ1yhwJbxi1DXxmWQuxttRmDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "+g7uZBMLADV28v3ebKfzsovu00t7yOky2RSlWr+Bjh8="
          },
          "tokens_per_usd_hour": 97
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama",
          "tokens_per_sec": 87.36,
          "cost_usd_per_million": 2.9697,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "g7D+Q6k6zJhxTSi+PBsK06ZcHIejAEnTvClNj43djf0L+uTIb2uN4qKfakhrxurLDQRxh6BkvlT7HDWhndFNDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "7/KuC28eu6TwKMyb+IJy6dUVh1xPI/0rSPXR1DOwQF0="
          },
          "tokens_per_usd_hour": 94
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "BF16 (16-bit, full precision) by unsloth",
          "build_page": "/inferences/qwen-3/4b/unsloth-bf16-ollama",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:BF16",
          "engine": "ollama",
          "tokens_per_sec": 80.76,
          "cost_usd_per_million": 3.2502,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 16384,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "Uw667P0enOb1mXsOG2PG/D1nlbskOfpiphmG8pZEDi22HTTrGW+/YGgcrlNVD7Xip4fE3LSpslMVVoXLiL4DCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "eaxgpNqS1poyhyGlh9rXFc2ru+qyC94yTwq3BYBIHfE="
          },
          "tokens_per_usd_hour": 85
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "Q4_K_M (4-bit) by maziyarpanahi",
          "build_page": null,
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 78.03,
          "cost_usd_per_million": 3.284,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "u2VgH5kp3E/E7Yl5Y3fWzcYTUrWFr+lqo5EVgo1z9aqtg1niQIlOT5SHK21M0X44te90Pq0D0BQpefWNIYN0CQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "vKBZGRzGOo1PUiaS5jFBPhzIW2tOen8wY4lEWWyKlpc="
          },
          "tokens_per_usd_hour": 85
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q4_K_S (4-bit) by maziyarpanahi",
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q4_k_s-ollama",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q4_K_S",
          "engine": "ollama",
          "tokens_per_sec": 77.39,
          "cost_usd_per_million": 3.3111,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "t2qkAy0C+sPNIaON6cHhKeEnjsLTaGPE9z2zw2jVUkN/32v8vxPx1OzeKT9iOOv18fFV2cV8Srb1HWgWMEDxAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "DiQp5zspuoHVy+qH2LoVx2ntYmh68NMEDHIsXcwF5Dw="
          },
          "tokens_per_usd_hour": 84
        },
        {
          "model": "Llama 3.2 3B",
          "page": "/inferences/llama-3-2/3b",
          "build": "Q8_0 (8-bit) by unsloth",
          "build_page": "/inferences/llama-3-2/3b/unsloth-q8_0-ollama",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q8_0",
          "engine": "ollama",
          "tokens_per_sec": 72.81,
          "cost_usd_per_million": 3.5559,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "go91DeLQ6idWbHdt213XRm0c8EDfjU/0nfTfS9aSqeP0kF58LdDFSzbMS64CeU/SA3K+ElxpdTIrtLJub/GmDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "+YAK7oWyWELAwxYOFPwnqJCzagwu2SjDvJz2qz7IN/E="
          },
          "tokens_per_usd_hour": 78
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "IQ3_XXS (3-bit) by unsloth",
          "build_page": "/inferences/qwen-3/4b/unsloth-iq3_xxs-ollama",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ3_XXS",
          "engine": "ollama",
          "tokens_per_sec": 53.3,
          "cost_usd_per_million": 4.8319,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "fqpSLr0DIktLWUNRbhlVmB6jo7kvBLgjwB4c4g0enMP59zsysO0/+5uxNCO9TnjCgP4ZRNqD6mQPOYVrOkHwDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "1HzlZGbLI9BFmVGxRIFsTOaPNX72b7Ex1ImkgOlRrg0="
          },
          "tokens_per_usd_hour": 57
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q8_0 (8-bit) by maziyarpanahi",
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q8_0-ollama",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q8_0",
          "engine": "ollama",
          "tokens_per_sec": 39.21,
          "cost_usd_per_million": 6.5351,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "0/2WQviYqJwPbo3QvSX1sNfzcdaUjb4qQwthVAbyyAJW71qosWL3uULyYAPHSguF1FEmyyT2ATCPRpHMpza4AA==",
            "signing_payload_b64": "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",
            "payload_sha256": "A3Q9puLABfdtMMlZiXzlvOnfSQvJQ2xUQGFbYgZnO3I="
          },
          "tokens_per_usd_hour": 43
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "Q5_K_M (5-bit) by Qwen",
          "build_page": "/inferences/qwen-3/4b/qwen-q5_k_m-ollama",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_K_M",
          "engine": "ollama",
          "tokens_per_sec": 35.37,
          "cost_usd_per_million": 7.3075,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "lgGV/xIplFUolwGepCLHVGUlT+0S5Hv+r4WK1+51tz9WWJB5kf1osV+9iM0LsGKMHfV84/zhD+rKzSWoMsgpBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "YUrgQW/fZIKjq9JaNS8Wm++Mnb6Kh5p13il9R2Zco3c="
          },
          "tokens_per_usd_hour": 38
        },
        {
          "model": "Sarvam-1",
          "page": "/inferences/sarvam/1",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 19.76,
          "cost_usd_per_million": 13.0304,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "F5Jbiivz/n2bnD4rajHRfruggRLe1yf+AS4X33N+kCSsurC0wAHLRvFgacwSWdSb4tIVK1hSp1jeA4zh2k6QCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "xr717MF2lwiKoiUdiQo17Th06gp/cYr9zTA0dstiJc8="
          },
          "tokens_per_usd_hour": 21
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "Q8_0 (8-bit) by Qwen",
          "build_page": "/inferences/qwen-3/4b/qwen-q8_0-ollama",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q8_0",
          "engine": "ollama",
          "tokens_per_sec": 17.06,
          "cost_usd_per_million": 15.2173,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "7aM93YA3I1eAMZcXrta9FpVV8P1YxiPE7vX1eoVlPIu9cv7G7DBuZ8q/7O8pa19lN3YcipEgOage8Q+XIOQ1Aw==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTcuMDU3NzEwODE4NDY2NDQsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0MzA5NjUyOTk1MDMyMTAwLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwYmM2NTBiNTlhZjg3OTQ4OTYwMzFkMmU4MmJjYTBlMTE5ZjM5MGUyMWQ2OTA5NjNhZTQ3NTc4MjM4MDg0YmY1NCIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoid2luZG93cy9hbWQ2NC9udmlkaWEvMTZnIiwibW9kZWwiOiJoZi5jby9Rd2VuL1F3ZW4zLTRCLUdHVUY6UThfMCIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYSIsInNhbXBsZV9jb3VudCI6MTMsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6MTUwLjc4MjI0ODk5OTk5OTk4LCJ0b3RhbF90b2tlbnMiOjI1NzJ9fQ==",
            "payload_sha256": "IPNDr5TRCITlgsQvUlEUsG4vx9i0s0flLE5H0PzBFjg="
          },
          "tokens_per_usd_hour": 18
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "Q5_0 (5-bit) by Qwen",
          "build_page": "/inferences/qwen-3/4b/qwen-q5_0-ollama",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_0",
          "engine": "ollama",
          "tokens_per_sec": 11.16,
          "cost_usd_per_million": 23.165,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "gKGVC0w0zcnjCGiEOMVNFVOuXbAKFoTznWkIPByFIKhnuxcRUORqTnAXvtVJCsk2K/+S7nV8YQNeV7ph4A5cAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "q6B2aQ0b7pYORXtDd1wLJBK/ROa/BnF4xMI7b2CJk0M="
          },
          "tokens_per_usd_hour": 12
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "BF16 (16-bit, full precision) by unsloth",
          "build_page": "/inferences/phi-4/mini/unsloth-bf16-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:BF16",
          "engine": "ollama",
          "tokens_per_sec": 0.08,
          "cost_usd_per_million": 3352.4625,
          "measured_at": "2026-11-10T11:20:52.995Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "amgW31bT4QjODZ4OEX0ZaMqxLjh7Fd3u67gXsufUi48FYCmK/SpuC7+DJX2ihMbBuPLbtMTtNAijp5ho1ZdPAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "ZX/46eYWosd48NGvUtPQfF5DsyMGVyi5EhzDbLbqe1A="
          },
          "tokens_per_usd_hour": 0
        }
      ],
      "rank": 5,
      "of": 36,
      "percentiles": {
        "gpu_tflops": 80,
        "memory_bandwidth": 89,
        "cpu_all_cores": 96,
        "cpu_one_core": 96,
        "disk_read": 11,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "e3654f",
      "did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
      "name": "Gram ·e3654f",
      "claimed": false,
      "online": false,
      "last_seen": "2026-10-30T10:49:46.768Z",
      "version": "07f247b",
      "region": {
        "code": "NL_9711",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2.2,
        "cpus": 4,
        "ram_gb": 31.346599578857422,
        "memory_bandwidth_gb_s": 0.2,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 2983.7,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
          "clock_ghz": 2.2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 256,
          "l3_mb": 55,
          "single_core_hashes_per_sec": 140.6,
          "multi_core_hashes_per_sec": 167.5,
          "sustained_ratio": 0.5,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-10-29T23:59:50.606Z",
          "measured_in": "NL_9711",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 0.2,
          "latency_ns": 161.5,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-10-29T23:59:51.651Z",
          "measured_in": "NL_9711",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 2983.7,
          "write_mb_s": 1.5,
          "random_read_iops": 281724,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-10-30T00:00:37.531Z",
          "measured_in": "NL_9711",
          "signed": true
        }
      },
      "signed_runs": 49,
      "signed_runs_30d": 23,
      "builds_measured": 49,
      "best_results": 19,
      "first_signed_at": "2026-10-28T07:45:00.331Z",
      "last_signed_at": "2026-10-30T06:35:24.096Z",
      "awake_pct": 100,
      "queue": {
        "measured": 23,
        "failed": 13,
        "queued": 1355
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-29",
          "hours": 5
        },
        {
          "date": "2026-10-30",
          "hours": 11
        }
      ],
      "uptime_pct": 33.3,
      "best_value": {
        "model": "Forge-1-Mini",
        "page": null,
        "build": null,
        "build_page": "/inferences/north-1/forge-1-mini~north-ml1/north-ml1-q2_k-ollama",
        "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q2_K",
        "engine": "ollama-host",
        "tokens_per_sec": 1074.92,
        "cost_usd_per_million": 0.0283,
        "measured_at": "2026-10-30T06:35:24.096Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "IrkEGr17iVZaj/ODrXupMsMKUHTnFJ1OBbe3wkcu745GdsFBTtKF/ay4Y7zgxgUX1dNyj9V9F5gyT4rXR8FuDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "CukUdyh0rDsTNqukQVnwOjBK7dhlbqriuHCqE73ThiI="
        },
        "tokens_per_usd_hour": 9817
      },
      "opened": [
        {
          "ref": "claude-haiku-4-5",
          "family": null,
          "member": null,
          "at": "2026-10-30T05:04:36.371Z"
        },
        {
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_M",
          "family": "gpt-oss",
          "member": "20b",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3.6-27B-GGUF:Q4_K_M",
          "family": "qwen-3-6",
          "member": "27b",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/meta-models/Muse-Glimmer-30B-GGUF:Q4_K_M",
          "family": "muse-glimmer",
          "member": "30b",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q3_K_S",
          "family": "gpt-oss",
          "member": "20b",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_0",
          "family": "gpt-oss",
          "member": "20b",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/bartowski/moonshotai_Kimi-Linear-48B-A3B-Instruct-GGUF:Q4_K_M",
          "family": "kimi-linear",
          "member": "48b-a3b",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/huihui-ai/Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF:Q6_K",
          "family": "qwen-3-6",
          "member": "35b-a3b~huihui-ai-abliterated-mtp",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/tencent/Hy-MT2-30B-A3B-GGUF:Q4_K_M",
          "family": "hy-mt-2",
          "member": "30b-a3b",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q2_K",
          "family": "gpt-oss",
          "member": "20b",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q8_0",
          "family": "gpt-oss",
          "member": "20b",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/unsloth/North-Mini-Code-1.0-GGUF:Q4_K_M",
          "family": "north-1",
          "member": "mini-code-1-0",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/bartowski/openai_gpt-oss-20b-GGUF-MXFP4-Experimental:MXFP4",
          "family": "gpt-oss",
          "member": "20b~bartowski-openai-experimental",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/DavidAU/OpenAi-GPT-oss-20b-abliterated-uncensored-NEO-Imatrix-gguf:Q8_0",
          "family": "gpt-oss",
          "member": "20b~huihui-ai-abliterated",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/DavidAU/OpenAi-GPT-oss-20b-HERETIC-uncensored-NEO-Imatrix-gguf:Q8_0",
          "family": "gpt-oss",
          "member": "20b~p-e-w-heretic",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q5_K_M",
          "family": "gpt-oss",
          "member": "20b~arliai-derestricted",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/mradermacher/Kimi-Linear-48B-A3B-Base-GGUF:Q4_K_M",
          "family": "kimi-linear",
          "member": "48b-a3b-base",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q3_K_S",
          "family": "gpt-oss",
          "member": "20b~davidau-openai-claude-4-5-opus-heretic-uncensored",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q4_0",
          "family": "gpt-oss",
          "member": "20b~davidau-openai-claude-4-5-opus-heretic-uncensored",
          "at": "2026-10-30T06:35:24.096Z"
        },
        {
          "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q2_K",
          "family": "north-1",
          "member": "forge-1-mini~north-ml1",
          "at": "2026-10-30T06:35:24.096Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-28",
          "runs": 14
        },
        {
          "date": "2026-10-29",
          "runs": 352
        },
        {
          "date": "2026-10-30",
          "runs": 470
        }
      ],
      "runs_by_family": [
        {
          "family": "mms-tts",
          "runs": 12
        },
        {
          "family": "gpt-oss",
          "runs": 11
        },
        {
          "family": "whisper-indic-ft",
          "runs": 3
        },
        {
          "family": "nllb",
          "runs": 2
        },
        {
          "family": "moonshine",
          "runs": 2
        },
        {
          "family": "qwen-3-6",
          "runs": 2
        },
        {
          "family": "kimi-linear",
          "runs": 2
        },
        {
          "family": "north-1",
          "runs": 2
        },
        {
          "family": "llama-3-2",
          "runs": 1
        },
        {
          "family": "glm-4-7",
          "runs": 1
        },
        {
          "family": "other",
          "runs": 1
        },
        {
          "family": "kokoro",
          "runs": 1
        },
        {
          "family": "f5-tts",
          "runs": 1
        },
        {
          "family": "muse-glimmer",
          "runs": 1
        },
        {
          "family": "whisper",
          "runs": 1
        },
        {
          "family": "wespeaker",
          "runs": 1
        },
        {
          "family": "ecapa-tdnn",
          "runs": 1
        },
        {
          "family": "madlad",
          "runs": 1
        },
        {
          "family": "hy-mt-2",
          "runs": 1
        },
        {
          "family": "qwen-3-5",
          "runs": 1
        },
        {
          "family": "sarvam-translate",
          "runs": 1
        }
      ],
      "rooms": [
        "NL_9711",
        "US_29415",
        "US_97058"
      ],
      "awards": [
        {
          "id": "runs-10",
          "label": "10+ signed runs"
        },
        {
          "id": "opener",
          "label": "First to measure 20 builds"
        },
        {
          "id": "best-10",
          "label": "Best result on 19 builds"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "Forge-1-Mini",
        "page": null,
        "build": null,
        "build_page": "/inferences/north-1/forge-1-mini~north-ml1/north-ml1-q2_k-ollama",
        "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q2_K",
        "engine": "ollama-host",
        "tokens_per_sec": 1074.92,
        "cost_usd_per_million": 0.0283,
        "measured_at": "2026-10-30T06:35:24.096Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "IrkEGr17iVZaj/ODrXupMsMKUHTnFJ1OBbe3wkcu745GdsFBTtKF/ay4Y7zgxgUX1dNyj9V9F5gyT4rXR8FuDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "CukUdyh0rDsTNqukQVnwOjBK7dhlbqriuHCqE73ThiI="
        },
        "tokens_per_usd_hour": 9817
      },
      "cheapest": {
        "model": "Forge-1-Mini",
        "page": null,
        "build": null,
        "build_page": "/inferences/north-1/forge-1-mini~north-ml1/north-ml1-q2_k-ollama",
        "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q2_K",
        "engine": "ollama-host",
        "tokens_per_sec": 1074.92,
        "cost_usd_per_million": 0.0283,
        "measured_at": "2026-10-30T06:35:24.096Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "IrkEGr17iVZaj/ODrXupMsMKUHTnFJ1OBbe3wkcu745GdsFBTtKF/ay4Y7zgxgUX1dNyj9V9F5gyT4rXR8FuDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "CukUdyh0rDsTNqukQVnwOjBK7dhlbqriuHCqE73ThiI="
        },
        "tokens_per_usd_hour": 9817
      },
      "builds": [
        {
          "model": "Forge-1-Mini",
          "page": null,
          "build": null,
          "build_page": "/inferences/north-1/forge-1-mini~north-ml1/north-ml1-q2_k-ollama",
          "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q2_K",
          "engine": "ollama-host",
          "tokens_per_sec": 1074.92,
          "cost_usd_per_million": 0.0283,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "IrkEGr17iVZaj/ODrXupMsMKUHTnFJ1OBbe3wkcu745GdsFBTtKF/ay4Y7zgxgUX1dNyj9V9F5gyT4rXR8FuDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "CukUdyh0rDsTNqukQVnwOjBK7dhlbqriuHCqE73ThiI="
          },
          "tokens_per_usd_hour": 9817
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": null,
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "tokens_per_sec": 11.45,
          "cost_usd_per_million": 2.6566,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "VwsbNGAF8UjFxH5JNbn4FCMEzEO4kR9zn//+9cHV5sEi0nm5UvPBDwURrb4XElNl1cc5+12lUWkUOJsX0y2aCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "qrUwcerxoHklzLicmUtNKQ8zcVGaGQaMF19KWfgt82o="
          },
          "tokens_per_usd_hour": 105
        },
        {
          "model": "Hunyuan MT 2 30B-A3B",
          "page": "/inferences/hy-mt-2/30b-a3b",
          "build": "Q4_K_M (4-bit) by tencent",
          "build_page": "/inferences/hy-mt-2/30b-a3b/tencent-q4_k_m-ollama",
          "ref": "hf.co/tencent/Hy-MT2-30B-A3B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 5.36,
          "cost_usd_per_million": 8.3129,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "RTFgNEfiOHEhYFohoGoxFFyKell+w3kWlYqqiVggODJk9UopRoPwCwPSPfOtJV2qp4j5pZnjeRgHnIcpU0LeCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "MntXlcKKYxYiFJcxaBqL2uG+JQGUu5LAZo85uJSjD3Y="
          },
          "tokens_per_usd_hour": 33
        },
        {
          "model": "OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
          "page": "/inferences/gpt-oss/20b~davidau-openai-claude-4-5-opus-heretic-uncensored",
          "build": "Q4_0 (4-bit) by mradermacher",
          "build_page": "/inferences/gpt-oss/20b~davidau-openai-claude-4-5-opus-heretic-uncensored/mradermacher-q4_0-ollama",
          "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q4_0",
          "engine": "ollama-host",
          "tokens_per_sec": 4.82,
          "cost_usd_per_million": 8.2607,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "tWuYKHXMN171qhfeC2n94lPtEPX0hy10zLDTcSbRgECJD1Wzf9lArjimkKPZSQQnEz8ni5VIrWLzenyulRc6BQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "uRLhHieWlXAf5UsGL3B7oejF/sRsQgtSGXwHmITGL+s="
          },
          "tokens_per_usd_hour": 34
        },
        {
          "model": "Kimi Linear 48B-A3B",
          "page": "/inferences/kimi-linear/48b-a3b",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/kimi-linear/48b-a3b/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/moonshotai_Kimi-Linear-48B-A3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 4.81,
          "cost_usd_per_million": 11.1806,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "YoELUsrRXI23U/VSXa434Rs3mzV4HdxRxj5kFzfivKCYDZLhtsLqnLZ1a8LlQsgSq1lCJ0CAZd82m9FAFNTfDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "Fus3Neuar8wsWRwwszugDZ3VoBe/dlQ3FA4A63ERgZo="
          },
          "tokens_per_usd_hour": 25
        },
        {
          "model": "OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
          "page": "/inferences/gpt-oss/20b~davidau-openai-claude-4-5-opus-heretic-uncensored",
          "build": "Q3_K_S (3-bit) by mradermacher",
          "build_page": "/inferences/gpt-oss/20b~davidau-openai-claude-4-5-opus-heretic-uncensored/mradermacher-q3_k_s-ollama",
          "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 4.72,
          "cost_usd_per_million": 8.4252,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "+xtVIsd6qc1eivlsb6wbHAbvpew5wzmWuw4WWotRENaG9xyUZ8sWhrfa4KCYs5CVk+sBDLZaWtTbvwFz+6gNDQ==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NC43MjEyMTQ3MDY3NzY5NCwidW5pdCI6InRva2Vucy9zIiwidHMiOjE3OTMzNDIxMjQwOTYwMDQ3NTgsIm5vZGVfZGlkIjoiZGlkOmVwbjowMDI0MDgwMTEyMjBhNDg0ZjhhMGEzZTQyNzI2YTM4ZTFmMTU3N2Y1YmVlNTMwZTUxODU3MTg0YTgxYmFlODk2NzYzNWQ2ZTM2NTRmIiwiZXh0cmEiOnsiaGFyZHdhcmVfY2xhc3MiOiJsaW51eC9hbWQ2NC9ub25lLzMyZyIsIm1vZGVsIjoiaGYuY28vbXJhZGVybWFjaGVyL09wZW5BSS1ncHQtb3NzLTIwQi1DbGF1ZGUtNC41LU9wdXMtSGVyZXRpYy1VbmNlbnNvcmVkLWkxLUdHVUY6UTNfS19TIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjUzLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjM1NS4yMDUxOTY5OTk5OTk5NCwidG90YWxfdG9rZW5zIjoxNjc3fX0=",
            "payload_sha256": "SkIjIbR7PzFwQbcV/Ng/nk8P0dAOxVdHZBYodBYhJ6k="
          },
          "tokens_per_usd_hour": 33
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q4_0 (4-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q4_0-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_0",
          "engine": "ollama-host",
          "tokens_per_sec": 4.56,
          "cost_usd_per_million": 8.6295,
          "measured_at": "2026-11-14T20:22:46.948Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "YyzbZTOjhiM+NIhCD3X67fna6xFNzI6VTHGQZmRY5zqb6y1sSFwAXFlW05zBBwC2U2cAsQy6pb3KGri4hlWVCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "CfyUetf0TNjuJqVUDuD7MgL2vm6jU9oWa4MsYSUBfv4="
          },
          "tokens_per_usd_hour": 32
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 4.41,
          "cost_usd_per_million": 8.9491,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "+G+ybq6tBWwm8EeOJZCgzn79CShJvim8ErkVPpzvObHOUhDPdB/KA3o9E7uGP6DiI32t9OYZET9XE/9EhybnCA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NC40MDcwMTM2NDQ5NjYyODksInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzMzQyMTI0MDk2MDA0NzU4LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwYTQ4NGY4YTBhM2U0MjcyNmEzOGUxZjE1NzdmNWJlZTUzMGU1MTg1NzE4NGE4MWJhZTg5Njc2MzVkNmUzNjU0ZiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbm9uZS8zMmciLCJtb2RlbCI6ImhmLmNvL3Vuc2xvdGgvZ3B0LW9zcy0yMGItR0dVRjpRNF9LX00iLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6NDIsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6MTYwMC44NTczMDc5OTk5OTk2LCJ0b3RhbF90b2tlbnMiOjcwNTV9fQ==",
            "payload_sha256": "0R0oWGjp4U8rA980GTXB+sz8mG6Wr/4Ig/X4ubJEbIc="
          },
          "tokens_per_usd_hour": 31
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q3_K_S (3-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q3_k_s-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 4.35,
          "cost_usd_per_million": 9.032,
          "measured_at": "2026-11-15T06:44:44.996Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "F0s6Gh5rDuyr4vEDOeR1kqpH1W+yQ1M/KH/J6iKXQs29UGm/mSrfN47JCnAyS5hxolalU0U9RZH9rjzTd/GyBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "x/e4uK04T8R7jWxRm8p5HcufuBBn/lOtxOOeyjHRSI4="
          },
          "tokens_per_usd_hour": 31
        },
        {
          "model": "openai_gpt-oss-20b-GGUF-MXFP4-Experimental",
          "page": "/inferences/gpt-oss/20b~bartowski-openai-experimental",
          "build": "MXFP4 (4-bit) by bartowski",
          "build_page": "/inferences/gpt-oss/20b~bartowski-openai-experimental/bartowski-mxfp4-ollama",
          "ref": "hf.co/bartowski/openai_gpt-oss-20b-GGUF-MXFP4-Experimental:MXFP4",
          "engine": "ollama-host",
          "tokens_per_sec": 4.32,
          "cost_usd_per_million": 9.2234,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "ONqpEZ/YbmMbLPLlvr6GytD+vdXFcuCBgeLje1HSOvKuHBTCxHqvcT+bXrzjxoQVKn3OREeZ9RJ+FNMA2bfUDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "SXkhE2immoEOD7ZrlPMOMqIQe9WSF+He+zOFn6OEDcE="
          },
          "tokens_per_usd_hour": 30
        },
        {
          "model": "Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF",
          "page": "/inferences/qwen-3-6/35b-a3b~huihui-ai-abliterated-mtp",
          "build": "Q6_K (6-bit) by huihui-ai",
          "build_page": "/inferences/qwen-3-6/35b-a3b~huihui-ai-abliterated-mtp/huihui-ai-q6_k-ollama",
          "ref": "hf.co/huihui-ai/Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF:Q6_K",
          "engine": "ollama-host",
          "tokens_per_sec": 4.29,
          "cost_usd_per_million": 12.3819,
          "measured_at": "2026-11-14T20:22:46.948Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "xt9ES5v2myeNju0C0/oQI786YQUEMxKxNpPpbPmWpzzTRJnOHbS7qby1zi/guxzZNOh+j2ad+ED3z+Y8pOVPCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "A7asW5Anl42qeNaOTyfVt5iKgX98VQhbPu/Ahb7aZ+k="
          },
          "tokens_per_usd_hour": 22
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q2_K (2-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q2_k-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q2_K",
          "engine": "ollama-host",
          "tokens_per_sec": 4.16,
          "cost_usd_per_million": 9.4507,
          "measured_at": "2026-11-15T06:44:44.996Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "RKQoJp5fVa+d3jp+CBOHgEaNM9UvrXNUuyjlcWyMO+/FeAH2Fgy0mBtD7BbaEm+DjSOG8Wbuqhl62ta43+WJDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "PUqF+vuaVswGiydPi/94HmxVrP1bGe4m3jIO/DhOaNI="
          },
          "tokens_per_usd_hour": 29
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q8_0 (8-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q8_0-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 4.04,
          "cost_usd_per_million": 9.856,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "gi+quH7r/k3h8tXvHURih+O0v8Tf50UIU4u1VlA9rESv5+aX1UyiHHUqhu7c7O0vsxemXusW0clM3VGWg9PbBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "F1WzqmDROeCS7UKDO3BgqNkwhlmvZDMBhI7bj2slhgI="
          },
          "tokens_per_usd_hour": 28
        },
        {
          "model": "GLM-4.7 Flash",
          "page": "/inferences/glm-4-7/flash",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/glm-4-7/flash/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/GLM-4.7-Flash-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 3.55,
          "cost_usd_per_million": 12.5701,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "xGfHrry32UfdIWxUCLLWOfeG3HgDHy04wN3bxNtJbQAlnvDAGvkMyU7Ylr7eRz0kwhHxyVWo93aRzVURcEogBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "M+qXLEAZFbxdEpXxUK7hawQRcB1vxQyNmAGjYWGeEf0="
          },
          "tokens_per_usd_hour": 22
        },
        {
          "model": "gpt-oss-20b-Derestricted",
          "page": "/inferences/gpt-oss/20b~arliai-derestricted",
          "build": "Q5_K_M (5-bit) by MaziyarPanahi",
          "build_page": "/inferences/gpt-oss/20b~arliai-derestricted/maziyarpanahi-q5_k_m-ollama",
          "ref": "hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q5_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 3.44,
          "cost_usd_per_million": 12.6617,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "wjZhPCq87FwRwC3fOzwkRRxkttNBG+RuwgmnpOjmtmcCjlXlHeQW7GmZFAZFV7nL7Zl14KQp3KT7vGRzqSySAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "wWbY3d/Hjew+8RmG7ErFXuDBvttPkje/+j7czbXyeOc="
          },
          "tokens_per_usd_hour": 22
        },
        {
          "model": "gpt-oss-20b-heretic",
          "page": "/inferences/gpt-oss/20b~p-e-w-heretic",
          "build": "Q8_0 (8-bit) by DavidAU",
          "build_page": "/inferences/gpt-oss/20b~p-e-w-heretic/davidau-q8_0-ollama",
          "ref": "hf.co/DavidAU/OpenAi-GPT-oss-20b-HERETIC-uncensored-NEO-Imatrix-gguf:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 3.36,
          "cost_usd_per_million": 14.1515,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "9CtCK+QPb2aORsG3/0sopimYsJBzsnrgxjjHiSMZwvT3ePHGAaQUA4d9JwfvVwWu0VeLXUvc+vRtFzu+P12qBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "GkcVC3nJaiPx/dAJIT/SlZlwhZ7LJ/mXI2FGqOej6jI="
          },
          "tokens_per_usd_hour": 20
        },
        {
          "model": "Huihui-gpt-oss-20b-BF16-abliterated",
          "page": "/inferences/gpt-oss/20b~huihui-ai-abliterated",
          "build": "Q8_0 (8-bit) by DavidAU",
          "build_page": "/inferences/gpt-oss/20b~huihui-ai-abliterated/davidau-q8_0-ollama",
          "ref": "hf.co/DavidAU/OpenAi-GPT-oss-20b-abliterated-uncensored-NEO-Imatrix-gguf:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 3.28,
          "cost_usd_per_million": 14.5051,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "7fq/nfUOIwWb1MEiE8f3ep48tfNxf1eW2P5+nkOEhpuA0tZQdKVuXMOOMHXxtCbew5+WpXV9nail1XqzYZArBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "f7DPvL366THj5jr8nPyomgXT9C+aDUS1tA6+H5oQaB8="
          },
          "tokens_per_usd_hour": 19
        },
        {
          "model": "North 1.0 Mini-Code-1-0",
          "page": "/inferences/north-1/mini-code-1-0",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/north-1/mini-code-1-0/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/North-Mini-Code-1.0-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 3.2,
          "cost_usd_per_million": 14.1577,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "4NB1Ukny6rhc1gY5u7446P3vFgG6oB9moRtcbcbWfww4UJOlNqUWfOT/66qELlEB+cwYmROmGNEnDD8wVrq1Cw==",
            "signing_payload_b64": "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",
            "payload_sha256": "RUJWzDwzT1IdZg+m2HmpV1AKoMLpujT0k3JPev+qVZc="
          },
          "tokens_per_usd_hour": 20
        },
        {
          "model": "Qwen 3.5 9B",
          "page": "/inferences/qwen-3-5/9b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/qwen-3-5/9b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Qwen3.5-9B-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 2.39,
          "cost_usd_per_million": null,
          "measured_at": "2026-10-30T04:00:04.405Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "hW44MIAlhpIPtlh9xuVlT/Dpy0a0MTg5/IgLpjdl8QkdYc3DNhGfreArnIAUjTVTFDj119t4qsBiLeusr8oZDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "oz8IG1hK8gawyLDYnAqOzN4JM0bBpyglTJi6n/FlKFc="
          },
          "tokens_per_usd_hour": null
        }
      ],
      "rank": 6,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 4,
        "cpu_all_cores": 11,
        "cpu_one_core": 21,
        "disk_read": 50,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "2e5c02",
      "did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
      "name": "Gram ·2e5c02",
      "claimed": true,
      "online": false,
      "last_seen": "2026-11-15T22:48:28.191Z",
      "version": "46aa233",
      "region": {
        "code": "IN_500001",
        "city": "Hyderabad",
        "state": "Telangana"
      },
      "hardware": {
        "os": "windows",
        "arch": "x86_64",
        "class": "windows/amd64/intel/16g",
        "cpu_model": "11th Gen Intel Core i5-1135G7",
        "cpu_clock_ghz": null,
        "cpus": 8,
        "ram_gb": 15.788383483886719,
        "memory_bandwidth_gb_s": 1.2,
        "gpu": "Intel(R) Iris(R) Xe Graphics",
        "vram_gb": 1,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 329.7,
        "gpu_count": 1,
        "gpu_org": "intel",
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "11th Gen Intel Core i5-1135G7",
          "model_raw": "11th Gen Intel Core i5-1135G7",
          "clock_ghz": null,
          "cores_logical": 8,
          "cores_physical": null,
          "threads_per_core": null,
          "sockets": null,
          "l2_kb": null,
          "l3_mb": null,
          "single_core_hashes_per_sec": 502.2,
          "multi_core_hashes_per_sec": 1190.3,
          "sustained_ratio": 0.9,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-15T12:56:42.595Z",
          "measured_in": "IN_500001",
          "signed": true
        },
        "memory": {
          "total_gb": 15.8,
          "bandwidth_gb_s": 1.2,
          "latency_ns": 253.1,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-15T12:56:43.806Z",
          "measured_in": "IN_500001",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 329.7,
          "write_mb_s": 12.4,
          "random_read_iops": 8193,
          "fs_type": null,
          "device_model": null,
          "measured_at": "2026-11-15T12:56:49.437Z",
          "measured_in": "IN_500001",
          "signed": true
        }
      },
      "signed_runs": 46,
      "signed_runs_30d": 32,
      "builds_measured": 46,
      "best_results": 20,
      "first_signed_at": "2026-10-23T14:59:04.241Z",
      "last_signed_at": "2026-11-15T14:04:06.883Z",
      "awake_pct": 100,
      "queue": {
        "measured": 21,
        "failed": 37,
        "queued": 1559
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-30",
          "hours": 4
        },
        {
          "date": "2026-11-05",
          "hours": 7
        },
        {
          "date": "2026-11-06",
          "hours": 11
        },
        {
          "date": "2026-11-07",
          "hours": 12
        },
        {
          "date": "2026-11-08",
          "hours": 9
        },
        {
          "date": "2026-11-09",
          "hours": 7
        },
        {
          "date": "2026-11-10",
          "hours": 11
        },
        {
          "date": "2026-11-11",
          "hours": 8
        },
        {
          "date": "2026-11-12",
          "hours": 8
        },
        {
          "date": "2026-11-13",
          "hours": 18
        },
        {
          "date": "2026-11-14",
          "hours": 18
        },
        {
          "date": "2026-11-15",
          "hours": 8
        }
      ],
      "uptime_pct": 42,
      "best_value": {
        "model": "Gemma 3 270M",
        "page": "/inferences/gemma-3/270m",
        "build": "IQ2_M (2-bit) by unsloth",
        "build_page": "/inferences/gemma-3/270m/unsloth-iq2_m-ollama",
        "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_M",
        "engine": "ollama",
        "tokens_per_sec": 68.95,
        "cost_usd_per_million": 1.0314,
        "measured_at": "2026-11-15T14:04:06.882Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
          "signature": "9G+CFgOxofbjtvP7XJaOsiscKDpQv5maZUbkMdsMYyApDt7kF7YpXIjzIV9V+PiiTUBP7sUL7DaaUjtOYPlVAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "HmvwBeRg3+rEpGBWlrg50P1+q+7bt6atl5pj0nUNLFo="
        },
        "tokens_per_usd_hour": 269
      },
      "opened": [
        {
          "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q2_K",
          "family": "glm-ocr",
          "member": "base",
          "at": "2026-11-15T14:04:06.883Z"
        },
        {
          "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q3_K_S",
          "family": "glm-ocr",
          "member": "base",
          "at": "2026-11-15T14:04:06.883Z"
        },
        {
          "ref": "hf.co/mradermacher/Hy-MT2-1.8B-GGUF:Q2_K",
          "family": "hy-mt-2",
          "member": "1.8b",
          "at": "2026-11-15T14:04:06.883Z"
        },
        {
          "ref": "hf.co/mradermacher/Hy-MT2-1.8B-GGUF:Q3_K_S",
          "family": "hy-mt-2",
          "member": "1.8b",
          "at": "2026-11-15T14:04:06.883Z"
        },
        {
          "ref": "hf.co/mradermacher/MiniCPM-V-4.6-gguf:Q2_K",
          "family": "minicpm-v-4-6",
          "member": "base",
          "at": "2026-11-15T14:04:06.883Z"
        },
        {
          "ref": "hf.co/mradermacher/MiniCPM4.1-8B-Russian-CPT-i1-GGUF:IQ3_XXS",
          "family": "minicpm-4-1",
          "member": "8b",
          "at": "2026-11-15T14:04:06.883Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-23",
          "runs": 6
        },
        {
          "date": "2026-10-30",
          "runs": 46
        },
        {
          "date": "2026-11-05",
          "runs": 42
        },
        {
          "date": "2026-11-06",
          "runs": 30
        },
        {
          "date": "2026-11-07",
          "runs": 209
        },
        {
          "date": "2026-11-08",
          "runs": 138
        },
        {
          "date": "2026-11-09",
          "runs": 165
        },
        {
          "date": "2026-11-10",
          "runs": 237
        },
        {
          "date": "2026-11-11",
          "runs": 240
        },
        {
          "date": "2026-11-12",
          "runs": 92
        },
        {
          "date": "2026-11-13",
          "runs": 322
        },
        {
          "date": "2026-11-14",
          "runs": 666
        },
        {
          "date": "2026-11-15",
          "runs": 198
        }
      ],
      "runs_by_family": [
        {
          "family": "gemma-3",
          "runs": 4
        },
        {
          "family": "embeddinggemma",
          "runs": 3
        },
        {
          "family": "hy-mt-2",
          "runs": 3
        },
        {
          "family": "functiongemma",
          "runs": 3
        },
        {
          "family": "glm-ocr",
          "runs": 3
        },
        {
          "family": "phi-4",
          "runs": 2
        },
        {
          "family": "depth-anything",
          "runs": 2
        },
        {
          "family": "minicpm-v-4-6",
          "runs": 2
        },
        {
          "family": "sd-turbo",
          "runs": 2
        },
        {
          "family": "llama-3-2",
          "runs": 1
        },
        {
          "family": "musicgen",
          "runs": 1
        },
        {
          "family": "gemma-4",
          "runs": 1
        },
        {
          "family": "sam",
          "runs": 1
        },
        {
          "family": "qwen-2-5",
          "runs": 1
        },
        {
          "family": "nllb",
          "runs": 1
        },
        {
          "family": "sarvam",
          "runs": 1
        },
        {
          "family": "tiny-sd",
          "runs": 1
        },
        {
          "family": "mms-tts",
          "runs": 1
        },
        {
          "family": "f5-tts",
          "runs": 1
        },
        {
          "family": "minicpm-5",
          "runs": 1
        },
        {
          "family": "seamless-m4t",
          "runs": 1
        },
        {
          "family": "granite-4-2",
          "runs": 1
        },
        {
          "family": "ministral-3",
          "runs": 1
        },
        {
          "family": "qwen-3",
          "runs": 1
        },
        {
          "family": "minicpm-v-4",
          "runs": 1
        },
        {
          "family": "deepseek-ocr-2",
          "runs": 1
        },
        {
          "family": "translategemma",
          "runs": 1
        },
        {
          "family": "deepseek-ocr",
          "runs": 1
        },
        {
          "family": "hy-mt-1-5",
          "runs": 1
        },
        {
          "family": "qwen-3-5",
          "runs": 1
        },
        {
          "family": "minicpm-4-1",
          "runs": 1
        }
      ],
      "rooms": [
        "21GVTmGuZJZ6SY9tENvf",
        "asPepgKbRT1dsKNkRGZd",
        "aWBauzUULQa1A60adrwc",
        "ecfmNiLz2x7vWNymNbtq",
        "EUTOPIA_FOUNDATION",
        "IN_500001",
        "j9p5QqSCECG55k3Qf98Z",
        "mavmNeRcYj0L06b4Sgiw",
        "oh1PHVetYJiodEZRRqoS",
        "P9GypLlx75EXirEjOgkl",
        "rSqKUZuufpk7scsG9rGP",
        "uxloWJEkaJ1rf77ezuyN",
        "WnsRXXyOFYej2nEm4kSg"
      ],
      "awards": [
        {
          "id": "runs-10",
          "label": "10+ signed runs"
        },
        {
          "id": "opener",
          "label": "First to measure 6 builds"
        },
        {
          "id": "best-10",
          "label": "Best result on 20 builds"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "Gemma 3 270M",
        "page": "/inferences/gemma-3/270m",
        "build": "IQ2_M (2-bit) by unsloth",
        "build_page": "/inferences/gemma-3/270m/unsloth-iq2_m-ollama",
        "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_M",
        "engine": "ollama",
        "tokens_per_sec": 68.95,
        "cost_usd_per_million": 1.0314,
        "measured_at": "2026-11-15T14:04:06.882Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
          "signature": "9G+CFgOxofbjtvP7XJaOsiscKDpQv5maZUbkMdsMYyApDt7kF7YpXIjzIV9V+PiiTUBP7sUL7DaaUjtOYPlVAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "HmvwBeRg3+rEpGBWlrg50P1+q+7bt6atl5pj0nUNLFo="
        },
        "tokens_per_usd_hour": 269
      },
      "cheapest": {
        "model": "Gemma 3 270M",
        "page": "/inferences/gemma-3/270m",
        "build": "IQ2_M (2-bit) by unsloth",
        "build_page": "/inferences/gemma-3/270m/unsloth-iq2_m-ollama",
        "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_M",
        "engine": "ollama",
        "tokens_per_sec": 68.95,
        "cost_usd_per_million": 1.0314,
        "measured_at": "2026-11-15T14:04:06.882Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
          "signature": "9G+CFgOxofbjtvP7XJaOsiscKDpQv5maZUbkMdsMYyApDt7kF7YpXIjzIV9V+PiiTUBP7sUL7DaaUjtOYPlVAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "HmvwBeRg3+rEpGBWlrg50P1+q+7bt6atl5pj0nUNLFo="
        },
        "tokens_per_usd_hour": 269
      },
      "builds": [
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "IQ2_M (2-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-iq2_m-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_M",
          "engine": "ollama",
          "tokens_per_sec": 68.95,
          "cost_usd_per_million": 1.0314,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "9G+CFgOxofbjtvP7XJaOsiscKDpQv5maZUbkMdsMYyApDt7kF7YpXIjzIV9V+PiiTUBP7sUL7DaaUjtOYPlVAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "HmvwBeRg3+rEpGBWlrg50P1+q+7bt6atl5pj0nUNLFo="
          },
          "tokens_per_usd_hour": 269
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "IQ2_XXS (2-bit) by unsloth",
          "build_page": "/inferences/functiongemma/270m/unsloth-iq2_xxs-ollama",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:IQ2_XXS",
          "engine": "ollama",
          "tokens_per_sec": 66.76,
          "cost_usd_per_million": 1.0652,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "aqDiwuUniKURoR8YnkaDL6q78AStiinzUVA0ABckw67aFt7NwzaJID6LC0+46RQEy8DUe8fMGxChQXqMprolDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "d+rby3WMXs+DyBdUeVYzEVrw/+yzeGSjcAu+8wqaY20="
          },
          "tokens_per_usd_hour": 261
        },
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "IQ3_XXS (3-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-iq3_xxs-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ3_XXS",
          "engine": "ollama",
          "tokens_per_sec": 52.33,
          "cost_usd_per_million": 1.3589,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "/1h1wIBUh4G0+vn8FkW/uITxL+O0cvTgB/uH5SVX+DRWG051m/qlM9dnwIGRBXfpbLlZjxWlxEwMMAamp2plCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "5Buy1YIHoXmtncx3nB+f5xn/+OBc9XyDTkPHB5N5kjE="
          },
          "tokens_per_usd_hour": 204
        },
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 41.45,
          "cost_usd_per_million": 1.717,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "LogKNw4iHnjAhtX6trv+yW4zW50T6yMOsxPBRmTg2WFXJYo7JjXJM05a5M9D5COC6kwvcHaY1vBFa6sKx5qyBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "hnv3vpJ0FPXAT/elYqji8cFZffMzzhL26Rkpbp7BchA="
          },
          "tokens_per_usd_hour": 162
        },
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "IQ2_XXS (2-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-iq2_xxs-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_XXS",
          "engine": "ollama",
          "tokens_per_sec": 34.6,
          "cost_usd_per_million": 2.0551,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "/Oz4CzLFCswmhw6znYJCNBw7jG7Au4+JDS64ADkWcjOStvTXT1WpuDpUGIWcELyQdZ8tInEPi626Wh88hlglAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "DgCqjlr+cP8mdeFOqh2MkfP0G0OSNhBNJn/OSfh2z6E="
          },
          "tokens_per_usd_hour": 135
        },
        {
          "model": "MiniCPM-V 4.6 Base",
          "page": "/inferences/minicpm-v-4-6/base",
          "build": "Q4_K_M (4-bit) by openbmb",
          "build_page": "/inferences/minicpm-v-4-6/base/openbmb-q4_k_m-ollama",
          "ref": "hf.co/openbmb/MiniCPM-V-4.6-gguf:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 27.29,
          "cost_usd_per_million": 2.6154,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "ZM26p9MeA5fgyWiG7k4BdWSDBgnPv+UxTwnE58tCPg5qKf9ug/s9VIrYNdEp+sjQf2xO03K48UX7+krIw9y6CQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "eOoDCzjxkJBjVnKripjXgZuVBgyAVx+9wyhhmja5ID8="
          },
          "tokens_per_usd_hour": 106
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/functiongemma/270m/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 23.66,
          "cost_usd_per_million": 3.0082,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "jvSCo5FLY3tF4kRI/BluRZV2pZnl1EqCYwXzHoEr4w5dv4vVnm51L1E4ut8+mqNu2GJ8y0bPim5/h0v/FlNNAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "ehDr29HCZkLDu8m0JS7mbVw6XxFvdLSVb5p/QA7uHB8="
          },
          "tokens_per_usd_hour": 92
        },
        {
          "model": "MiniCPM5 1B",
          "page": "/inferences/minicpm-5/1b",
          "build": "Q4_K_M (4-bit) by openbmb",
          "build_page": "/inferences/minicpm-5/1b/openbmb-q4_k_m-ollama",
          "ref": "hf.co/openbmb/MiniCPM5-1B-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 20.58,
          "cost_usd_per_million": 3.475,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "TIaWwHklgMQcq9MaF4I+hUe94cKZonVPoNNbjD4oU32l6Uy8LN9HZ0up9gDuQDnFEi1t3ATcgh1dbvwTJk6cBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "fwi+MDyQRHe+7NL9pYSzGNYC6RpRtz6LaVFxLXmL7nA="
          },
          "tokens_per_usd_hour": 80
        },
        {
          "model": "GLM-OCR Base",
          "page": "/inferences/glm-ocr/base",
          "build": "Q3_K_S (3-bit) by mradermacher",
          "build_page": "/inferences/glm-ocr/base/mradermacher-q3_k_s-ollama",
          "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q3_K_S",
          "engine": "ollama",
          "tokens_per_sec": 18.11,
          "cost_usd_per_million": 3.9367,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "ZsKCFSMr5DcBu/B1ixxi5QwKtsWPLgIp01mije4h3cq/kIsNepOrWOE2ILkEGX1HKAe3I51VpsQ4blRDu8qVBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "pmu2r4qiir7/gUGI9yTh1kVmbCabMNuUgwAehq+ZveE="
          },
          "tokens_per_usd_hour": 71
        },
        {
          "model": "DeepSeek-OCR 2 Base",
          "page": "/inferences/deepseek-ocr-2/base",
          "build": "Q4_K_M (4-bit) by sabafallah",
          "build_page": "/inferences/deepseek-ocr-2/base/sabafallah-q4_k_m-ollama",
          "ref": "hf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 15.36,
          "cost_usd_per_million": 4.7185,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "t3MoE3dafF90/XXpeHCShcVUq8ypYrk9B9sWbGrtUDa0hwG3KqHZSLhYWiIA0NNGawXcr5u45AOKyhJUhrAzBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "UcmVc07/VVPZHH94nQgV99MmfbBugE3CbBOHsYPY5+c="
          },
          "tokens_per_usd_hour": 59
        },
        {
          "model": "Hunyuan MT 2 1.8B",
          "page": "/inferences/hy-mt-2/1.8b",
          "build": "Q3_K_S (3-bit) by mradermacher",
          "build_page": "/inferences/hy-mt-2/1.8b/mradermacher-q3_k_s-ollama",
          "ref": "hf.co/mradermacher/Hy-MT2-1.8B-GGUF:Q3_K_S",
          "engine": "ollama",
          "tokens_per_sec": 15.31,
          "cost_usd_per_million": 4.6786,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "/ujE9QwdaKcDrlf7aFO5HlWFJaxhotaa3I7uS8wP6qQ0a5JZfbGznK76W/3h376YNiR2h06JxCDzjIQ124J7Cw==",
            "signing_payload_b64": "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",
            "payload_sha256": "zcN+6TqviVRG5jGJ0pkRxVAuaM9oZSplJH7Uv8JqMcc="
          },
          "tokens_per_usd_hour": 59
        },
        {
          "model": "Hunyuan MT 2 1.8B",
          "page": "/inferences/hy-mt-2/1.8b",
          "build": "Q4_K_M (4-bit) by tencent",
          "build_page": "/inferences/hy-mt-2/1.8b/tencent-q4_k_m-ollama",
          "ref": "hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 14.17,
          "cost_usd_per_million": 5.0706,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "Cghz/vOSgpMIIJTmK4/7JxW09UynwtcpE1/7PM5nruGcm4x3wLJQViHt6iwbWriJ2sLLyFxbg0qZl0d9QmaVBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "j+HXRQHEjnhv0tvpkFb5wLZ8N8ZnG7pzUj+wDrVNEUQ="
          },
          "tokens_per_usd_hour": 55
        },
        {
          "model": "DeepSeek-OCR Base",
          "page": "/inferences/deepseek-ocr/base",
          "build": "Q4_K_M (4-bit) by sabafallah",
          "build_page": "/inferences/deepseek-ocr/base/sabafallah-q4_k_m-ollama",
          "ref": "hf.co/sabafallah/DeepSeek-OCR-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 13.32,
          "cost_usd_per_million": 5.4435,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "fCcoLkBF4l/BECmDHM40HrCL2mPJpkNSdr1QYDyv4WzGJhAcA1kPBvHOg21NFNWy2Uycu3DRh/BhZ9ku0vt3Bg==",
            "signing_payload_b64": "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",
            "payload_sha256": "2wwrJBEw+e8137zc3dLRhDSekaQbSJZ59qFU7z2ksxM="
          },
          "tokens_per_usd_hour": 51
        },
        {
          "model": "GLM-OCR Base",
          "page": "/inferences/glm-ocr/base",
          "build": "Q2_K (2-bit) by mradermacher",
          "build_page": "/inferences/glm-ocr/base/mradermacher-q2_k-ollama",
          "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q2_K",
          "engine": "ollama",
          "tokens_per_sec": 13.16,
          "cost_usd_per_million": 5.4152,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "uMeGKAvkpkRs1TByJvc65FAWKE+HEk6J9zdLau1lQmVe2iysO5y5JS3QjCiI6ZMS9b2j4EwDMj1eml32dYzyDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "4d+ShCANdBvwSyApV2hlm4WtSFnP8hGhysotsgqECBA="
          },
          "tokens_per_usd_hour": 51
        },
        {
          "model": "GLM-OCR Base",
          "page": "/inferences/glm-ocr/base",
          "build": "Q4_K_M (4-bit) by mradermacher",
          "build_page": "/inferences/glm-ocr/base/mradermacher-q4_k_m-ollama",
          "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 12.98,
          "cost_usd_per_million": 5.502,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "1rxnp5E1PvRGeRpVM6osmnrm+ZmNyJ2BUVlmED2i1SbXA1prrFXO9CnIAEsHLnwITXVcc759MDhWmCsVTSblDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "KkcoFM5gxEif40CODU5thjSAebUomuS0+Jsiaxrshrw="
          },
          "tokens_per_usd_hour": 51
        },
        {
          "model": "Hunyuan MT 1.5 1.8B",
          "page": "/inferences/hy-mt-1-5/1.8b",
          "build": "Q4_K_M (4-bit) by tencent",
          "build_page": "/inferences/hy-mt-1-5/1.8b/tencent-q4_k_m-ollama",
          "ref": "hf.co/tencent/HY-MT1.5-1.8B-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 7.41,
          "cost_usd_per_million": 9.6986,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "AL8OpM9y/9FbdJu2l6PLLBZKKS3fwAnSI/ySdbvmJ4hQbrVYqvTHKliA0r2AL3yHLzvQW75vpqcJR2/mofNbDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "QAui5ZrGmqP3ObE5U+lcBKsLmWgW6Nte9XTifeNVPUI="
          },
          "tokens_per_usd_hour": 29
        },
        {
          "model": "Sarvam-1",
          "page": "/inferences/sarvam/1",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 4.83,
          "cost_usd_per_million": 14.9543,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "zOzcOojwFSJa7iY6P5I80gNY8jssi+BFdjn17lLOoOPsjgao35PnumPcrLTVuE3LxO1137PaHWvuL5NKQbygCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "c3rFbcnYJtjbMbmxSYoj1nDeHBECdXtsM+T8bXwVrOM="
          },
          "tokens_per_usd_hour": 19
        },
        {
          "model": "Ministral 3 3B-Instruct-2512",
          "page": "/inferences/ministral-3/3b-instruct-2512",
          "build": "Q4_K_M (4-bit) by mistralai",
          "build_page": "/inferences/ministral-3/3b-instruct-2512/mistralai-q4_k_m-ollama",
          "ref": "hf.co/mistralai/Ministral-3-3B-Instruct-2512-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 4.66,
          "cost_usd_per_million": 15.5973,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "ykNQ80DXwnULjVCK1xBzggzo9FiwauQOKZkox0ur+/MQcvLjFIYmSOC8ec0qlLLx6QDe0C76GH3QeSe/SwI7AQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "uEEtDl6YSJa7MhAkqO5u4jPkUP8DiRMAl9Fi3OT+BGM="
          },
          "tokens_per_usd_hour": 18
        },
        {
          "model": "MiniCPM-V 4 Base",
          "page": "/inferences/minicpm-v-4/base",
          "build": "Q4_K_M (4-bit) by openbmb",
          "build_page": "/inferences/minicpm-v-4/base/openbmb-q4_k_m-ollama",
          "ref": "hf.co/openbmb/MiniCPM-V-4-gguf:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 3.21,
          "cost_usd_per_million": 22.6155,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "FBPpsFx6elyeHQSgjNz2YqfZeKDPdQmLtMwsDBzpz5uXHnSbkxmX6bHyffYi9cRpAEyJYrGl1/N+tkB9bOvpBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "S/lbERe6+UhTzqbM5Wndafz4ZUsdRC4MrTfglEf4q5U="
          },
          "tokens_per_usd_hour": 12
        },
        {
          "model": "Qwen3-VL-8B-Instruct-abliterated-v1",
          "page": "/inferences/qwen-3/vl-8b~prithivmlmods-abliterated",
          "build": "Q2_K (2-bit) by mradermacher",
          "build_page": "/inferences/qwen-3/vl-8b~prithivmlmods-abliterated/mradermacher-q2_k-ollama",
          "ref": "hf.co/mradermacher/Qwen3-VL-8B-Instruct-abliterated-GGUF:Q2_K",
          "engine": "ollama",
          "tokens_per_sec": 2.6,
          "cost_usd_per_million": 28.2271,
          "measured_at": "2026-11-15T14:04:06.882Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "VLTjDZqtspOKwEDQxQiPMtltKpIBt+S1H4XCnXXpuaoQpmSM2jGsC6T9m7daITKkkQKE3T9r0Sh302ECBXF7AA==",
            "signing_payload_b64": "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",
            "payload_sha256": "Py21e92GPkYTUpiSGvFf4B1o/bqmKCJsUWaM/3SbKls="
          },
          "tokens_per_usd_hour": 10
        }
      ],
      "rank": 7,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 22,
        "cpu_all_cores": 64,
        "cpu_one_core": 71,
        "disk_read": 7,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "acbe75",
      "did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
      "name": "Gram ·acbe75",
      "claimed": false,
      "online": false,
      "last_seen": "2026-10-25T07:55:25.960Z",
      "version": "6d1b153",
      "region": null,
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": null,
        "cpu_clock_ghz": null,
        "cpus": 4,
        "ram_gb": 31.346603393554688,
        "memory_bandwidth_gb_s": null,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": null,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": null
      },
      "sheet": null,
      "signed_runs": 45,
      "signed_runs_30d": 45,
      "builds_measured": 45,
      "best_results": 43,
      "first_signed_at": "2026-10-25T07:41:14.641Z",
      "last_signed_at": "2026-10-25T07:41:14.641Z",
      "awake_pct": 100,
      "queue": {
        "measured": 47,
        "failed": 5,
        "queued": 1561
      },
      "download_mib_per_sec": 173.1,
      "uptime_days": [],
      "uptime_pct": null,
      "best_value": {
        "model": "FunctionGemma 270M",
        "page": "/inferences/functiongemma/270m",
        "build": "IQ4_NL (4-bit) by Mungert",
        "build_page": "/inferences/functiongemma/270m~mungert/mungert-iq4_nl-ollama",
        "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:IQ4_NL",
        "engine": "ollama-host",
        "tokens_per_sec": 40.24,
        "cost_usd_per_million": 0.7588,
        "measured_at": "2026-10-25T07:41:14.640Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
          "signature": "12rypP10l3lR7QSQU36TUIklHhKvho4HxHulpCLRay2ojtKf2C1dfuOS6EHfXpEnCRWcMcJ/9ZplRIrfn20tAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "1ZdtjM17nqHoFZkclvLkt5mdyx8tH8OKJBcTHttsLco="
        },
        "tokens_per_usd_hour": 366
      },
      "opened": [
        {
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
          "family": "gemma-3",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M",
          "family": "functiongemma",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/ggml-org/gemma-3-1b-it-GGUF:Q4_K_M",
          "family": "gemma-3",
          "member": "1b",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q4_K_M",
          "family": "granite-4-0",
          "member": "1b",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:IQ2_M",
          "family": "functiongemma",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:IQ2_XXS",
          "family": "functiongemma",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_M",
          "family": "gemma-3",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_XXS",
          "family": "gemma-3",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ3_XXS",
          "family": "gemma-3",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q4_K_M",
          "family": "qwen-3-5",
          "member": "0.8b",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3.5-2B-GGUF:Q4_K_M",
          "family": "qwen-3-5",
          "member": "2b",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:F16",
          "family": "functiongemma",
          "member": "270m~mingxilei-unsloth-ft-qat",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q4_K_M",
          "family": "functiongemma",
          "member": "270m~mingxilei-unsloth-ft-qat",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q5_K_M",
          "family": "functiongemma",
          "member": "270m~mingxilei-unsloth-ft-qat",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q8_0",
          "family": "functiongemma",
          "member": "270m~mingxilei-unsloth-ft-qat",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:F16",
          "family": "functiongemma",
          "member": "270m~muxodious-abominable-heresy",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:IQ4_XS",
          "family": "functiongemma",
          "member": "270m~muxodious-abominable-heresy",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q2_K",
          "family": "functiongemma",
          "member": "270m~muxodious-abominable-heresy",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q3_K_S",
          "family": "functiongemma",
          "member": "270m~muxodious-abominable-heresy",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q4_K_M",
          "family": "functiongemma",
          "member": "270m~muxodious-abominable-heresy",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q4_K_S",
          "family": "functiongemma",
          "member": "270m~muxodious-abominable-heresy",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q8_0",
          "family": "functiongemma",
          "member": "270m~muxodious-abominable-heresy",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:F16",
          "family": "functiongemma",
          "member": "270m~acon96-home",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:IQ4_XS",
          "family": "functiongemma",
          "member": "270m~acon96-home",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q2_K",
          "family": "functiongemma",
          "member": "270m~acon96-home",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q3_K_S",
          "family": "functiongemma",
          "member": "270m~acon96-home",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q4_K_M",
          "family": "functiongemma",
          "member": "270m~acon96-home",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q4_K_S",
          "family": "functiongemma",
          "member": "270m~acon96-home",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q8_0",
          "family": "functiongemma",
          "member": "270m~acon96-home",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:BF16",
          "family": "functiongemma",
          "member": "270m~mungert",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:IQ4_NL",
          "family": "functiongemma",
          "member": "270m~mungert",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q3_K_S",
          "family": "functiongemma",
          "member": "270m~mungert",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q4_1",
          "family": "functiongemma",
          "member": "270m~mungert",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q4_K_M",
          "family": "functiongemma",
          "member": "270m~mungert",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q8_0",
          "family": "functiongemma",
          "member": "270m~mungert",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:BF16",
          "family": "functiongemma",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:IQ3_XXS",
          "family": "functiongemma",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_0",
          "family": "functiongemma",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q8_0",
          "family": "functiongemma",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:F16",
          "family": "gemma-3",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_0",
          "family": "gemma-3",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q8_0",
          "family": "gemma-3",
          "member": "270m",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/unsloth/LFM2.5-1.2B-Instruct-GGUF:Q4_K_M",
          "family": "lfm-2-5",
          "member": "1.2b",
          "at": "2026-10-25T07:41:14.641Z"
        },
        {
          "ref": "hf.co/victor/functiongemma-agent-gguf:Q4_K_M",
          "family": "functiongemma",
          "member": "270m~unsloth",
          "at": "2026-10-25T07:41:14.641Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-25",
          "runs": 45
        }
      ],
      "runs_by_family": [
        {
          "family": "functiongemma",
          "runs": 32
        },
        {
          "family": "gemma-3",
          "runs": 8
        },
        {
          "family": "qwen-3-5",
          "runs": 2
        },
        {
          "family": "llama-3-2",
          "runs": 1
        },
        {
          "family": "granite-4-0",
          "runs": 1
        },
        {
          "family": "lfm-2-5",
          "runs": 1
        }
      ],
      "rooms": [
        "TW_110"
      ],
      "awards": [
        {
          "id": "runs-10",
          "label": "10+ signed runs"
        },
        {
          "id": "opener",
          "label": "First to measure 44 builds"
        },
        {
          "id": "best-10",
          "label": "Best result on 43 builds"
        }
      ],
      "fastest": {
        "model": "FunctionGemma 270M",
        "page": "/inferences/functiongemma/270m",
        "build": "IQ4_NL (4-bit) by Mungert",
        "build_page": "/inferences/functiongemma/270m~mungert/mungert-iq4_nl-ollama",
        "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:IQ4_NL",
        "engine": "ollama-host",
        "tokens_per_sec": 40.24,
        "cost_usd_per_million": 0.7588,
        "measured_at": "2026-10-25T07:41:14.640Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
          "signature": "12rypP10l3lR7QSQU36TUIklHhKvho4HxHulpCLRay2ojtKf2C1dfuOS6EHfXpEnCRWcMcJ/9ZplRIrfn20tAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "1ZdtjM17nqHoFZkclvLkt5mdyx8tH8OKJBcTHttsLco="
        },
        "tokens_per_usd_hour": 366
      },
      "cheapest": {
        "model": "FunctionGemma 270M",
        "page": "/inferences/functiongemma/270m",
        "build": "IQ4_NL (4-bit) by Mungert",
        "build_page": "/inferences/functiongemma/270m~mungert/mungert-iq4_nl-ollama",
        "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:IQ4_NL",
        "engine": "ollama-host",
        "tokens_per_sec": 40.24,
        "cost_usd_per_million": 0.7588,
        "measured_at": "2026-10-25T07:41:14.640Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
          "signature": "12rypP10l3lR7QSQU36TUIklHhKvho4HxHulpCLRay2ojtKf2C1dfuOS6EHfXpEnCRWcMcJ/9ZplRIrfn20tAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "1ZdtjM17nqHoFZkclvLkt5mdyx8tH8OKJBcTHttsLco="
        },
        "tokens_per_usd_hour": 366
      },
      "builds": [
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "IQ4_NL (4-bit) by Mungert",
          "build_page": "/inferences/functiongemma/270m~mungert/mungert-iq4_nl-ollama",
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:IQ4_NL",
          "engine": "ollama-host",
          "tokens_per_sec": 40.24,
          "cost_usd_per_million": 0.7588,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "12rypP10l3lR7QSQU36TUIklHhKvho4HxHulpCLRay2ojtKf2C1dfuOS6EHfXpEnCRWcMcJ/9ZplRIrfn20tAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "1ZdtjM17nqHoFZkclvLkt5mdyx8tH8OKJBcTHttsLco="
          },
          "tokens_per_usd_hour": 366
        },
        {
          "model": "functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
          "page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat",
          "build": "Q4_K_M (4-bit) by mingxilei",
          "build_page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat/mingxilei-q4_k_m-ollama",
          "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 39.3,
          "cost_usd_per_million": 0.779,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "X0c/TQdxAjPDBm5R//2Zbl/6sJrak/DTsQ7XkKjiH/cEFfgx2tGndoGR30Ae/jI1bntQ0mgWphCeCYwcTQoOAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "lABtwKNxFNzg0EYUCvyDy1MjMKnmpdtRxElG0SOOqr8="
          },
          "tokens_per_usd_hour": 357
        },
        {
          "model": "functiongemma-270m-it-abominable-heresy",
          "page": "/inferences/functiongemma/270m~muxodious-abominable-heresy",
          "build": "IQ4_XS (4-bit) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-iq4_xs-ollama",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:IQ4_XS",
          "engine": "ollama-host",
          "tokens_per_sec": 38.58,
          "cost_usd_per_million": 0.7932,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "VehGAESlKyA6SWp7bwkmA1SQy/4QDonX/oMf0e7u02XZoyAjE4B0hedi/VKTcb9MFH4iN9lGZ7VIGVlpyRkfCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "z31lTggVwG3jLbJV0JKv0sLn9srxyzn96tpRRvmUsDM="
          },
          "tokens_per_usd_hour": 350
        },
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 37.38,
          "cost_usd_per_million": 0.819,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "4Hd1CZ/Lu7kOHzU4Y01dgGPHZBY/pVn0K9rz71a+qYin2redirsNjdTJGProHQgzGttRM9dScpjWJeuBwiH1DA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MzcuMzc4NzEyMDQwODYzMjEsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkyOTE0MDc0NjQwODkyNjI3LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwMjY3ZWI1Zjg5ZDg4YmZlZDFmYjMzNzk4Yzk2ZDZlYTk2Mzk5ZjM5MGNjNjgxNWJhZWM3ZDljYjE3N2FjYmU3NSIsImV4dHJhIjp7Im1vZGVsIjoiaGYuY28vdW5zbG90aC9nZW1tYS0zLTI3MG0taXQtR0dVRjpRNF9LX00iLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6NDUsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6OTguMTMwNzIxOTk5OTk5OTksInRvdGFsX3Rva2VucyI6MzY2OH19",
            "payload_sha256": "UdbAtK7zYHWB3yzf4ksBncfsrzxy5AFfRdAV0TxxqOs="
          },
          "tokens_per_usd_hour": 339
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "Q4_0 (4-bit) by unsloth",
          "build_page": "/inferences/functiongemma/270m/unsloth-q4_0-ollama",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_0",
          "engine": "ollama-host",
          "tokens_per_sec": 37.25,
          "cost_usd_per_million": 0.8216,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "1Xr6ZK4G15aTY71zCNrHq0IwNDP7RglJc7tsOZK3S7HbWLN/JiPYrMxvWoK8hrOEYacRd1xU71H5qbxZhzn9CA==",
            "signing_payload_b64": "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",
            "payload_sha256": "GDlirxBJhckrZHD4rJNMTAp+z+GURfIutfWxrW8z3VA="
          },
          "tokens_per_usd_hour": 338
        },
        {
          "model": "functiongemma-270m-it-abominable-heresy",
          "page": "/inferences/functiongemma/270m~muxodious-abominable-heresy",
          "build": "Q3_K_S (3-bit) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-q3_k_s-ollama",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 36.72,
          "cost_usd_per_million": 0.8333,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "plpjGiafaRVNMDh0zQtM9KRUvvKd8W9e3xQH0/mYOBl9m/ApvqzPWVagy1Umc6wY7WVMLCng4icGIVcOIxe5CQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "TeczI9y63ptv5SjkZPefNFL0r0w+XF+jMQ1kKpGx3RA="
          },
          "tokens_per_usd_hour": 333
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "Q4_K_M (4-bit) by victor",
          "build_page": "/inferences/functiongemma/270m~unsloth/victor-q4_k_m-ollama",
          "ref": "hf.co/victor/functiongemma-agent-gguf:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 36.7,
          "cost_usd_per_million": 0.834,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "sdK4piV7QQ+QfjlKJLTTNukUsM+uXtRAeBzz8nPfLbd9icdJM8FkdTDIs8dlyCIt+JQF13PLxNH6nC/F6JCHCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "zjwMzhVd0JAHWa40xqRET+b9qUJ40PyJUopzdPcItVg="
          },
          "tokens_per_usd_hour": 333
        },
        {
          "model": "Home-FunctionGemma-270m",
          "page": "/inferences/functiongemma/270m~acon96-home",
          "build": "Q3_K_S (3-bit) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-q3_k_s-ollama",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 35.88,
          "cost_usd_per_million": 0.8528,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "tzbpkEe8EG6WTXyAokn+JdJAn7jzS6fBXLSuV0TfhtY/Dio6v6pJREuBYqfT9AbugH5leQalYkRAE4WSNmgoCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "XZyv4iHicMADbGT3gbg1ck/XYMKGk0y2YXEQujqstsM="
          },
          "tokens_per_usd_hour": 326
        },
        {
          "model": "functiongemma-270m-it-abominable-heresy",
          "page": "/inferences/functiongemma/270m~muxodious-abominable-heresy",
          "build": "Q2_K (2-bit) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-q2_k-ollama",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q2_K",
          "engine": "ollama-host",
          "tokens_per_sec": 35.11,
          "cost_usd_per_million": 0.8715,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "UQ+ol2lRsaufmBqgAO4vM7JFpGkQm/MgpZ/jo6qnpKEBata/5s7767LwoWf2yxr4aczYHJK8WqMC7RVb2PNGCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "iFQOgk3Bh/r3va5uy25ceka3dWnB+/eOtvnDHMu0yms="
          },
          "tokens_per_usd_hour": 319
        },
        {
          "model": "Home-FunctionGemma-270m",
          "page": "/inferences/functiongemma/270m~acon96-home",
          "build": "IQ4_XS (4-bit) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-iq4_xs-ollama",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:IQ4_XS",
          "engine": "ollama-host",
          "tokens_per_sec": 34.81,
          "cost_usd_per_million": 0.8793,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "O0nKhtOFxpfIpya9pFEDmJW3HP3sRYRUoUqyRHrxwKwXbxi7OjWpWPZnhDyd2cLiu70atOePegxrEZaFL4avDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "lF352rhXqEF9b9MPhqiULX+B+ooFb2rDscp0zDP07hc="
          },
          "tokens_per_usd_hour": 316
        },
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "IQ2_M (2-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-iq2_m-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_M",
          "engine": "ollama-host",
          "tokens_per_sec": 34.58,
          "cost_usd_per_million": 0.8836,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "Z9INea8tByqLpWMDkSTpSec2Tpr97QRKv2kVQENcIsDN8FFpyYLQmZ99akLda7ya/1mN/eQYkhaUSNHNF/U2Ag==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MzQuNTg0Mjg0MjAxNzE4NTcsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkyOTE0MDc0NjQwODkyNjI3LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwMjY3ZWI1Zjg5ZDg4YmZlZDFmYjMzNzk4Yzk2ZDZlYTk2Mzk5ZjM5MGNjNjgxNWJhZWM3ZDljYjE3N2FjYmU3NSIsImV4dHJhIjp7Im1vZGVsIjoiaGYuY28vdW5zbG90aC9nZW1tYS0zLTI3MG0taXQtR0dVRjpJUTJfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo0Nywic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjoxNjIuOTY0MTk0MDAwMDAwMDIsInRvdGFsX3Rva2VucyI6NTYzNn19",
            "payload_sha256": "lS1Yf+wdEKPa7A7s34G7ywsOwgCAYVk5Hqtq6n8QRRI="
          },
          "tokens_per_usd_hour": 314
        },
        {
          "model": "Home-FunctionGemma-270m",
          "page": "/inferences/functiongemma/270m~acon96-home",
          "build": "Q2_K (2-bit) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-q2_k-ollama",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q2_K",
          "engine": "ollama-host",
          "tokens_per_sec": 34.52,
          "cost_usd_per_million": 0.8864,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "NrQzePjdOBv88WWNWKAzYKVTos1rauqZ0NktiZPcu6vFH9gn0I2lsIvrQXVP+lbqKsV28VpHm8i24JZ8SVBfBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "6h1i0nRLzcSte/na3Tw3ePx7lxC85ruqzVHAOJpmQVU="
          },
          "tokens_per_usd_hour": 313
        },
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "IQ2_XXS (2-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-iq2_xxs-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_XXS",
          "engine": "ollama-host",
          "tokens_per_sec": 34.48,
          "cost_usd_per_million": 0.8863,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "nln3fof0ALK0dGfVfn4TNpH4sH153ckDAnIxAQ3iWKmKFDENXKemnOt6AGtf7YRtChTkzVFGM2xrupYIM5lgCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "4c7bbWMNMH/Ned5wuRTIV5YqCtNJKL0lhA2f338/SsA="
          },
          "tokens_per_usd_hour": 313
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "Q3_K_S (3-bit) by Mungert",
          "build_page": "/inferences/functiongemma/270m~mungert/mungert-q3_k_s-ollama",
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 34.43,
          "cost_usd_per_million": 0.8878,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "9qMQqd9D6z6uOwkHDEhTOWNv+FEOlBQriHP6KlSGeXTEi8em0grsT9GwiUvOhmk40y3IzyL9bYxPxSuQrbYeDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "XrhOQ9yDiiOns7V/FIMu3gyylMs/+tFAgWxq3RzJx2o="
          },
          "tokens_per_usd_hour": 313
        },
        {
          "model": "functiongemma-270m-it-abominable-heresy",
          "page": "/inferences/functiongemma/270m~muxodious-abominable-heresy",
          "build": "Q4_K_M (4-bit) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-q4_k_m-ollama",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 33.93,
          "cost_usd_per_million": 0.9024,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "Q5Snaad4r/moyQqiT8Hv8fy7KjrckFuZKmNFebyzVQs+UkOIFlVpE3dUQI9guNNb7qadZdwYOrwBifMsMrL/Cg==",
            "signing_payload_b64": "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",
            "payload_sha256": "jDusQ0/LL6wRXU1izuQQIcFC/asDVW6PFSLt/XmHCBA="
          },
          "tokens_per_usd_hour": 308
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/functiongemma/270m/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 33.83,
          "cost_usd_per_million": 0.9049,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "TqjB1okKtMfuq9sZOu4z9S/AVZ1cgwWGNnzkKtjTjEEeUk6wdGI2m7oTB+XkNSo3YZ02Pk3IMy28aLsCkshKDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "wdVXt8Aij+TYoK0GNFncL4QqU47ZqUrySoXnvNOkTME="
          },
          "tokens_per_usd_hour": 307
        },
        {
          "model": "functiongemma-270m-it-abominable-heresy",
          "page": "/inferences/functiongemma/270m~muxodious-abominable-heresy",
          "build": "Q4_K_S (4-bit) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-q4_k_s-ollama",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q4_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 33.74,
          "cost_usd_per_million": 0.9073,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "EDvxRWQIUV2cpvBnp2JGI2p6osMxeJPLb7ahxUWzwxWNY2nlKmhg5gyOV9lg3o0kjk81VrfYJaLrlXw/WY1ECQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "W462zWF6heg3huc5yl3WZ8B/EWJRuVJXTeEJZiRvETY="
          },
          "tokens_per_usd_hour": 306
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "Q4_1 (4-bit) by Mungert",
          "build_page": "/inferences/functiongemma/270m~mungert/mungert-q4_1-ollama",
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q4_1",
          "engine": "ollama-host",
          "tokens_per_sec": 33.51,
          "cost_usd_per_million": 0.912,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "pcJRXb0qh1NagvQv1GdgbZeSwy4KIW3cROIGdQp7uEmGjnI6Yfxkzfki2MQ9YvxXwLGpMHLP7tRjuyUcEVoqBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "728EVkpbzQsIck8f4FjVulAsSiR5ve/AndwqpCU+Adc="
          },
          "tokens_per_usd_hour": 305
        },
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "IQ3_XXS (3-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-iq3_xxs-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "tokens_per_sec": 33.41,
          "cost_usd_per_million": 0.9147,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "uL2ZwCfnlyAc/7vRh+ojNNuVAZORMQu5PJH12WYmdZSYqyKhz2V0t2q97999UmEsz+Va75g+XiPh7YBy06MIDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "S2vxWz9LesteHVhdc2E1D43oCbOsgkODdRVgbAd3OxA="
          },
          "tokens_per_usd_hour": 304
        },
        {
          "model": "Home-FunctionGemma-270m",
          "page": "/inferences/functiongemma/270m~acon96-home",
          "build": "Q4_K_M (4-bit) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-q4_k_m-ollama",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 33.28,
          "cost_usd_per_million": 0.9197,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "lkMyeXX1dWklFUlCSoYjkEUhPO+SUsO04bfFn1SB6p6GzmHNlqFBkVhTiny+X1GSr0o0lO806VXQrebsRT+zAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "lKPUcZUSmw9v3Bzid6+RRGPQEEmr3A6NveVI+7EmLFs="
          },
          "tokens_per_usd_hour": 302
        },
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "Q4_0 (4-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_0-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_0",
          "engine": "ollama-host",
          "tokens_per_sec": 33,
          "cost_usd_per_million": 0.9274,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "FTZFv80vKJMZD1WcVJ8ofu+ljRDHev7FLjlP4QdwyMTpd+EDAhhRv4XpejKcTq+1JC/jCfknhfbwRkA6wzQZCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "fAYSM1jXIZt26pxWTxa0u+A4RmkO3qtduPPIW6v151w="
          },
          "tokens_per_usd_hour": 299
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "Q8_0 (8-bit) by unsloth",
          "build_page": "/inferences/functiongemma/270m/unsloth-q8_0-ollama",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 32.9,
          "cost_usd_per_million": 0.9315,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "Tea2B/IZ9A5RGXeqG14BFAEjri4XAAV4lBCG3jrBmoWwtDwhIgBnlrb/hj6jHa2XmCfYD7gQPCo/B/+LBU/VBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "8GJYxGqZkmPjaTv3a7vX/jtRzNKE7/V1/JHeN7LDrfI="
          },
          "tokens_per_usd_hour": 298
        },
        {
          "model": "Home-FunctionGemma-270m",
          "page": "/inferences/functiongemma/270m~acon96-home",
          "build": "Q4_K_S (4-bit) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-q4_k_s-ollama",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q4_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 32.8,
          "cost_usd_per_million": 0.9333,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "cjHtAisgoqrEm3tsu0iqmZobRO944Vr+tc7/A+Davy13qoW2FmgSjyojOHQyAFdF+rx/BUej3wIzBBbgEHBjCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "4d76KNwnw82us4h95HUo/V2x8HrqqK9SB0xxd9fGGng="
          },
          "tokens_per_usd_hour": 298
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "Q4_K_M (4-bit) by Mungert",
          "build_page": "/inferences/functiongemma/270m~mungert/mungert-q4_k_m-ollama",
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 32.75,
          "cost_usd_per_million": 0.9347,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "jleFQ4bmkTut9BEqG024uX94IWTNbD2w3gOZ0tnkg/cuUtjA3+TiBWPVBLir5500ExVDd7/SWpC3TZw9NpCIAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "/hM3vUAwHMEfRGl2otrZOWOQxYQ648wR68cDrWKR9uE="
          },
          "tokens_per_usd_hour": 297
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "IQ3_XXS (3-bit) by unsloth",
          "build_page": "/inferences/functiongemma/270m/unsloth-iq3_xxs-ollama",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "tokens_per_sec": 32.63,
          "cost_usd_per_million": 0.9367,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "d4CMG9d0or8ou2f0tEr4nJABR0q31ojLYldwMA60ADcrMyE4nPuA8EFb2kIdye3NLJ6Tz+LoyLC9I2Ir9IfKCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "3IzfEQcWPe+PgogAfbl4At74MqumxY86mYuFOntJWhw="
          },
          "tokens_per_usd_hour": 297
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "IQ2_M (2-bit) by unsloth",
          "build_page": "/inferences/functiongemma/270m/unsloth-iq2_m-ollama",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:IQ2_M",
          "engine": "ollama-host",
          "tokens_per_sec": 32.5,
          "cost_usd_per_million": 0.9401,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "LltJZgqjdcAsDLFka4VTfaXG+H2oX78veE5b6ekNbj4UnzXmfD55xgLOj8bXAzRsDQHNdqFlgWtI3mv7CbYvCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "Ip0Y7iDisyedZcIBG1ug30CZ1pJjM744QT3LSDissJ4="
          },
          "tokens_per_usd_hour": 295
        },
        {
          "model": "functiongemma-270m-it-abominable-heresy",
          "page": "/inferences/functiongemma/270m~muxodious-abominable-heresy",
          "build": "Q8_0 (8-bit) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-q8_0-ollama",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 32.19,
          "cost_usd_per_million": 0.9519,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "5MzpNlm50+knSNKTzQ3QQgwa6cZwym/+YeudrZ44LylQEhzvEohIxsbkIEg1GzHNVBLXNDTpe0cwY7zgCpUeDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "SGMtarQiteaRDuDgR2K+Ei//6+tvE7g22klIabjDM2Q="
          },
          "tokens_per_usd_hour": 292
        },
        {
          "model": "Home-FunctionGemma-270m",
          "page": "/inferences/functiongemma/270m~acon96-home",
          "build": "Q8_0 (8-bit) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-q8_0-ollama",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 31.99,
          "cost_usd_per_million": 0.9578,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "JtIK2OreU3Sx4Z8uESr0dtJ6iCbJAsitkamJ5Vux8D+zvHTdU/4nIfEvxCWfVyKIQPfSqPURW6wae50Ge+S5BA==",
            "signing_payload_b64": "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",
            "payload_sha256": "ivGWW/hE5ieGh2DOMOyrPp+U30SzJEdCR5+GeKDIJVM="
          },
          "tokens_per_usd_hour": 290
        },
        {
          "model": "functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
          "page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat",
          "build": "Q5_K_M (5-bit) by mingxilei",
          "build_page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat/mingxilei-q5_k_m-ollama",
          "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q5_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 31.81,
          "cost_usd_per_million": 0.9624,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "d0z8CazX/Uw7H/Zr4kTP3hYof5QIOukPUxkjycUu1th2pnVI4aAjBA99ceF611iDgpktUt1P00wcCT6/ebEWDw==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MzEuODEzMTM4ODYwODA3NTUsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkyOTE0MDc0NjQwODkyNjI3LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwMjY3ZWI1Zjg5ZDg4YmZlZDFmYjMzNzk4Yzk2ZDZlYTk2Mzk5ZjM5MGNjNjgxNWJhZWM3ZDljYjE3N2FjYmU3NSIsImV4dHJhIjp7Im1vZGVsIjoiaGYuY28vbWluZ3hpbGVpL2Z1bmN0aW9uZ2VtbWEtMjcwbS1pdC11bnNsb3RoLWJuYi00Yml0LWZ0LVFBVC1HR1VGOlE1X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo1MCwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjoyMTQuNzU0MDM2OTk5OTk5OTMsInRvdGFsX3Rva2VucyI6NjgzMn19",
            "payload_sha256": "YTXsNroHaiXX71xymeYQoWUggSCD5QcqUGzIQWW3AI4="
          },
          "tokens_per_usd_hour": 289
        },
        {
          "model": "functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
          "page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat",
          "build": "Q8_0 (8-bit) by mingxilei",
          "build_page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat/mingxilei-q8_0-ollama",
          "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 31.48,
          "cost_usd_per_million": 0.9735,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "ZPxeA4Ox8tjWTaC2+qyBFHznn8oe/gl3FGT1FQGE4ON25ak71GUKuYBXooawqMlL0Pt/RKUTEU4Wx04zznVgAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "YfmBkQrAZePfjUBqozpBtrntyhkfk/x8A4i5KrD2hGY="
          },
          "tokens_per_usd_hour": 285
        },
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "Q8_0 (8-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-q8_0-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 30.86,
          "cost_usd_per_million": 0.993,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "lIWNsN60xFm9JhtL4leDzKdAYN9HzbDvdleNcJvhqMI3IV//XCwCLtmFz4G3mzmtOiZA90dk+UcWxMdln/TTCA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MzAuODU3NjIyNDY2NTI5OTE4LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5MjkxNDA3NDY0MDg5MjYyNywibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDI2N2ViNWY4OWQ4OGJmZWQxZmIzMzc5OGM5NmQ2ZWE5NjM5OWYzOTBjYzY4MTViYWVjN2Q5Y2IxNzdhY2JlNzUiLCJleHRyYSI6eyJtb2RlbCI6ImhmLmNvL3Vuc2xvdGgvZ2VtbWEtMy0yNzBtLWl0LUdHVUY6UThfMCIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo1Niwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjo3My4zNjkyMjk5OTk5OTk5NywidG90YWxfdG9rZW5zIjoyMjY0fX0=",
            "payload_sha256": "ItRcKZ631zh9KtJ2KdbLchUokU/bSpj+V7CHn+Le3GM="
          },
          "tokens_per_usd_hour": 280
        },
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "F16 (16-bit, full precision) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-f16-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:F16",
          "engine": "ollama-host",
          "tokens_per_sec": 23.52,
          "cost_usd_per_million": 1.3114,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": 16384,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "L6sfn2qL4akIScivTrr6c7xREiH+1qjCgpqGkwqOoQEmCs+yYs6zSzpdCJPb7nX5rgb12IlEez5EKSuvSbPUBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "rSJP825Ft1QwPah6P9ou8Avoogn0rR+U1gA3RD0T3RE="
          },
          "tokens_per_usd_hour": 212
        },
        {
          "model": "functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
          "page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat",
          "build": "F16 (16-bit, full precision) by mingxilei",
          "build_page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat/mingxilei-f16-ollama",
          "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:F16",
          "engine": "ollama-host",
          "tokens_per_sec": 23.13,
          "cost_usd_per_million": 1.3335,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "NqgdnMzZSS9xU892U8djThcSq1GosDCGZjuQgLv3pn19JAq2u+GXKJOVClpH8Fc1RjemXFM0086SU+62g/qpDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "l4jWnAmGi5Of+pax/Pcyp1gkBGlMzXP/0DnipKcLVrU="
          },
          "tokens_per_usd_hour": 208
        },
        {
          "model": "functiongemma-270m-it-abominable-heresy",
          "page": "/inferences/functiongemma/270m~muxodious-abominable-heresy",
          "build": "F16 (16-bit, full precision) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-f16-ollama",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:F16",
          "engine": "ollama-host",
          "tokens_per_sec": 22.71,
          "cost_usd_per_million": 1.3578,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "LKKAt3FY3XjwAbqzTWeFaCVnF9iGvxaFurGkfPcYxVDWEcBlDCh78UXPhs3EhzwWY2rfo44DHxGZQhNI0LRMBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "qSbHUDWpDPf/fN8j6sdSpvFeY/ZYgx0/S9oHUz0NSAo="
          },
          "tokens_per_usd_hour": 205
        },
        {
          "model": "Home-FunctionGemma-270m",
          "page": "/inferences/functiongemma/270m~acon96-home",
          "build": "F16 (16-bit, full precision) by mradermacher",
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-f16-ollama",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:F16",
          "engine": "ollama-host",
          "tokens_per_sec": 22.54,
          "cost_usd_per_million": 1.3679,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "Wx2jZyk3lqJbZmqy9a67YBKi2qqjNJhIFO/LAx/MVWLd7E+0K3RkKmUQ6bQsgUmeFyxt7VpfxtKilDd5+iLLBA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MjIuNTQ0NzE5MDA3MjYzNzE2LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5MjkxNDA3NDY0MDg5MjYyNywibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDI2N2ViNWY4OWQ4OGJmZWQxZmIzMzc5OGM5NmQ2ZWE5NjM5OWYzOTBjYzY4MTViYWVjN2Q5Y2IxNzdhY2JlNzUiLCJleHRyYSI6eyJtb2RlbCI6ImhmLmNvL21yYWRlcm1hY2hlci9Ib21lLUZ1bmN0aW9uR2VtbWEtMjcwbS1HR1VGOkYxNiIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo1NCwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjo2MDMuMDY4MDYxOTk3OTk5OCwidG90YWxfdG9rZW5zIjoxMzU5Nn19",
            "payload_sha256": "Ab89laejN8zQQKUJG/qSyjOL7BzD594fJHx604XpeNs="
          },
          "tokens_per_usd_hour": 203
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "Q8_0 (8-bit) by Mungert",
          "build_page": "/inferences/functiongemma/270m~mungert/mungert-q8_0-ollama",
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 22.34,
          "cost_usd_per_million": 1.3783,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "I0/JT4uWZGByqIiKOyCbttKuIDIrpwY4n5zL6rbtZ7CapLZSEqB0rinhcMqRraRQmbWzTmNl2YXz/BBLs5A2DQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "2xLRKMAk2QyFmDNw8bMAmSXivnd8DQOsMua+TxzxHQY="
          },
          "tokens_per_usd_hour": 201
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "BF16 (16-bit, full precision) by Mungert",
          "build_page": "/inferences/functiongemma/270m~mungert/mungert-bf16-ollama",
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:BF16",
          "engine": "ollama-host",
          "tokens_per_sec": 20.08,
          "cost_usd_per_million": 1.5355,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "RI2HqiJ/k+JkKZrbHt+J7s+cFmkz1fpXQ4b1VovtIN5k73Bs9Myi4y54qKwUybUJkKPbVSQHW8GAaca+H9GuDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "/EIt/v1FEtqR6Q9k8ui4/gcCs0ER6IsWkVEF+ZbLLfs="
          },
          "tokens_per_usd_hour": 181
        },
        {
          "model": "FunctionGemma 270M",
          "page": "/inferences/functiongemma/270m",
          "build": "BF16 (16-bit, full precision) by unsloth",
          "build_page": "/inferences/functiongemma/270m/unsloth-bf16-ollama",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:BF16",
          "engine": "ollama-host",
          "tokens_per_sec": 18.71,
          "cost_usd_per_million": 1.6483,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "FUseo8mhzweHoLuZC97cBuMNEATK8ae5ouOqaYniRJWqL9VJbByDhRW+gKWR+MUelLN8lfWv/e2zuuoc5Ft1Dw==",
            "signing_payload_b64": "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",
            "payload_sha256": "hRd/ZmBQTafxmCDri2ZDC9h+DRATq3Nddu1d4mThe8U="
          },
          "tokens_per_usd_hour": 169
        },
        {
          "model": "Qwen 3.5 0.8B",
          "page": "/inferences/qwen-3-5/0.8b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/qwen-3-5/0.8b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 16.1,
          "cost_usd_per_million": 1.9149,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "gpBFJYw0a12RTzr4EKF+YxLNRyADxiuwj3z07LFVxtLOedP8V7DQoKBj/ug55LirhvQkU+pFsWXefqxY3eaCCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "nnvBePZD6Tlgrf/l+r+C7g/kTd59hShv4VjF3yoC740="
          },
          "tokens_per_usd_hour": 145
        },
        {
          "model": "LFM2.5 1.2B",
          "page": "/inferences/lfm-2-5/1.2b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/lfm-2-5/1.2b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/LFM2.5-1.2B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 15.9,
          "cost_usd_per_million": 1.949,
          "measured_at": "2026-10-25T07:41:14.640Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "o23Daz021BkMs3r0r4bdmCB54kDgtX9N4EJYMyITJ7BqUP96ZRvWIRSWIRW7VbGFjmdHSTsDOQ1oe9Jof8oVBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "fhMb3SBYYz22VO574Jru3iZWyjlOrtcamdB2Xy3v3jk="
          },
          "tokens_per_usd_hour": 143
        },
        {
          "model": "Gemma 3 1B",
          "page": "/inferences/gemma-3/1b",
          "build": "Q4_K_M (4-bit) by ggml-org",
          "build_page": "/inferences/gemma-3/1b/ggml-org-q4_k_m-ollama",
          "ref": "hf.co/ggml-org/gemma-3-1b-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 13.98,
          "cost_usd_per_million": 2.2199,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "yP8sNk1aVG8ndDXE3hGWT1Lp8lzjKWxYX6FjqZf2L7l06/ZqPmc5po2O5nITnjxVhoBkid68D+gbE8zS0zxDCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "SGmJiRvuTNS7M/OJOAxNYjDLOspJvqZkROzm3E8XXD8="
          },
          "tokens_per_usd_hour": 125
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": null,
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "tokens_per_sec": 11.45,
          "cost_usd_per_million": 2.6566,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "VwsbNGAF8UjFxH5JNbn4FCMEzEO4kR9zn//+9cHV5sEi0nm5UvPBDwURrb4XElNl1cc5+12lUWkUOJsX0y2aCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "qrUwcerxoHklzLicmUtNKQ8zcVGaGQaMF19KWfgt82o="
          },
          "tokens_per_usd_hour": 105
        },
        {
          "model": "Granite 4.0 1B",
          "page": "/inferences/granite-4-0/1b",
          "build": "Q4_K_M (4-bit) by ibm-granite",
          "build_page": "/inferences/granite-4-0/1b/ibm-granite-q4_k_m-ollama",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 11.37,
          "cost_usd_per_million": 2.744,
          "measured_at": "2026-10-31T00:18:37.298Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "nqLXOqxDriaxROVdKEjJzh4w1AwdH4z9t9XISaALgYLFWL/lVJv/M/CrIMEy3TEDbtrarVVB0DKziBwXiteQDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "z76Hz4HoYUep4hYU8qw84UMdoj1i6aybsAqYhxCvglw="
          },
          "tokens_per_usd_hour": 101
        }
      ],
      "rank": 8,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": null,
        "cpu_all_cores": null,
        "cpu_one_core": null,
        "disk_read": null,
        "download": 100
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "afbb8a",
      "did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
      "name": "Gram ·afbb8a",
      "claimed": false,
      "online": false,
      "last_seen": "2026-11-02T11:47:39.454Z",
      "version": "efaaa4d",
      "region": {
        "code": "IN_110054",
        "city": "Delhi",
        "state": "Delhi"
      },
      "hardware": {
        "os": "darwin",
        "arch": "arm64",
        "class": "darwin/arm64/apple/16g",
        "cpu_model": "Apple M4",
        "cpu_clock_ghz": null,
        "cpus": 10,
        "ram_gb": 16,
        "memory_bandwidth_gb_s": 42.7,
        "gpu": "Apple M4",
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 13543.1,
        "gpu_count": 1,
        "gpu_org": "apple",
        "cpu_org": "apple"
      },
      "sheet": {
        "cpu": {
          "model": "Apple M4",
          "model_raw": "Apple M4",
          "clock_ghz": null,
          "cores_logical": 10,
          "cores_physical": 10,
          "threads_per_core": 1,
          "sockets": 1,
          "l2_kb": 4096,
          "l3_mb": null,
          "single_core_hashes_per_sec": 2850.2,
          "multi_core_hashes_per_sec": 22838.7,
          "sustained_ratio": 1.1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-02T08:29:50.654Z",
          "measured_in": "IN_110054",
          "signed": true
        },
        "memory": {
          "total_gb": 16,
          "bandwidth_gb_s": 42.7,
          "latency_ns": 6.5,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-02T08:29:51.180Z",
          "measured_in": "IN_110054",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 13543.1,
          "write_mb_s": 1696,
          "random_read_iops": 1733096,
          "fs_type": null,
          "device_model": null,
          "measured_at": "2026-11-02T08:29:51.427Z",
          "measured_in": "IN_110054",
          "signed": true
        }
      },
      "signed_runs": 43,
      "signed_runs_30d": 36,
      "builds_measured": 43,
      "best_results": 17,
      "first_signed_at": "2026-10-16T09:41:09.296Z",
      "last_signed_at": "2026-11-02T11:31:50.442Z",
      "awake_pct": 31.8,
      "queue": {
        "measured": 2,
        "failed": 7,
        "queued": 1310
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-11-02",
          "hours": 4
        }
      ],
      "uptime_pct": 16.7,
      "best_value": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": "Q4_K_M (4-bit) by maziyarpanahi",
        "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-q4_k_m-ollama",
        "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 113.27,
        "cost_usd_per_million": 0.8056,
        "measured_at": "2026-11-02T11:31:50.442Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
          "signature": "v8YfvttGpZh94elK5bm4mPYsZhsxopjsTz4SgFL6U06m8TRt5JHd/fVvwNAKdgjanGZnagDv1tHEPQ1DSTb/Ag==",
          "signing_payload_b64": "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",
          "payload_sha256": "uaDhtI2e7/M5CoNNcthDrIfB1IkMJcolt3e8NxLixKM="
        },
        "tokens_per_usd_hour": 345
      },
      "opened": [
        {
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ3_XXS",
          "family": "qwen-3",
          "member": "4b",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/bartowski/krutrim-ai-labs_Krutrim-2-instruct-GGUF:Q4_K_M",
          "family": "krutrim",
          "member": "2",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:IQ3_XS",
          "family": "llama-3-2",
          "member": "1b",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_S",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/mradermacher/Qwen3-VL-8B-Instruct-abliterated-GGUF:Q2_K",
          "family": "qwen-3",
          "member": "vl-8b~prithivmlmods-abliterated",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_0",
          "family": "qwen-3",
          "member": "4b",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q3_K_M",
          "family": "gemma-4",
          "member": "e2b",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_0",
          "family": "gemma-4",
          "member": "e2b",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q2_K",
          "family": "phi-4",
          "member": "mini-reasoning",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ1_S",
          "family": "qwen-3",
          "member": "4b",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ2_M",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/bartowski/sarvamai_sarvam-m-GGUF:Q4_K_M",
          "family": "sarvam",
          "member": "m",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
          "family": "llama-3-2",
          "member": "1b",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_NL",
          "family": "gemma-4",
          "member": "e2b",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3-VL-4B-Instruct-GGUF:Q4_1",
          "family": "qwen-3",
          "member": "4b",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-uncensored-GGUF:IQ3_XS",
          "family": "llama-3-2",
          "member": "3b~chuanli11-llama-3-2-uncensored",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ3_XXS",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ4_XS",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q3_K_L",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q3_K_S",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q4_0",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q4_K_M",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q8_0",
          "family": "phi-4",
          "member": "mini-reasoning",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/lmstudio-community/Phi-4-reasoning-plus-GGUF:Q6_K",
          "family": "phi-4",
          "member": "reasoning-plus",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-3B-Instruct-GGUF:IQ1_S",
          "family": "llama-3-2",
          "member": "3b",
          "at": "2026-11-02T11:31:50.442Z"
        },
        {
          "ref": "hf.co/unsloth/gemma-4-E4B-it-GGUF:Q3_K_M",
          "family": "gemma-4",
          "member": "e4b",
          "at": "2026-11-02T11:31:50.442Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-16",
          "runs": 35
        },
        {
          "date": "2026-11-02",
          "runs": 395
        }
      ],
      "runs_by_family": [
        {
          "family": "phi-4",
          "runs": 16
        },
        {
          "family": "llama-3-2",
          "runs": 7
        },
        {
          "family": "gemma-4",
          "runs": 6
        },
        {
          "family": "qwen-3",
          "runs": 6
        },
        {
          "family": "sarvam",
          "runs": 2
        },
        {
          "family": "musicgen",
          "runs": 1
        },
        {
          "family": "sam",
          "runs": 1
        },
        {
          "family": "tiny-sd",
          "runs": 1
        },
        {
          "family": "krutrim",
          "runs": 1
        },
        {
          "family": "wan-2",
          "runs": 1
        },
        {
          "family": "qwen-image",
          "runs": 1
        }
      ],
      "rooms": [],
      "awards": [
        {
          "id": "runs-10",
          "label": "10+ signed runs"
        },
        {
          "id": "opener",
          "label": "First to measure 26 builds"
        },
        {
          "id": "best-10",
          "label": "Best result on 17 builds"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": "Q4_K_M (4-bit) by maziyarpanahi",
        "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-q4_k_m-ollama",
        "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 113.27,
        "cost_usd_per_million": 0.8056,
        "measured_at": "2026-11-02T11:31:50.442Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
          "signature": "v8YfvttGpZh94elK5bm4mPYsZhsxopjsTz4SgFL6U06m8TRt5JHd/fVvwNAKdgjanGZnagDv1tHEPQ1DSTb/Ag==",
          "signing_payload_b64": "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",
          "payload_sha256": "uaDhtI2e7/M5CoNNcthDrIfB1IkMJcolt3e8NxLixKM="
        },
        "tokens_per_usd_hour": 345
      },
      "cheapest": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": "Q4_K_M (4-bit) by maziyarpanahi",
        "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-q4_k_m-ollama",
        "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 113.27,
        "cost_usd_per_million": 0.8056,
        "measured_at": "2026-11-02T11:31:50.442Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
          "signature": "v8YfvttGpZh94elK5bm4mPYsZhsxopjsTz4SgFL6U06m8TRt5JHd/fVvwNAKdgjanGZnagDv1tHEPQ1DSTb/Ag==",
          "signing_payload_b64": "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",
          "payload_sha256": "uaDhtI2e7/M5CoNNcthDrIfB1IkMJcolt3e8NxLixKM="
        },
        "tokens_per_usd_hour": 345
      },
      "builds": [
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": "Q4_K_M (4-bit) by maziyarpanahi",
          "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-q4_k_m-ollama",
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 113.27,
          "cost_usd_per_million": 0.8056,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "v8YfvttGpZh94elK5bm4mPYsZhsxopjsTz4SgFL6U06m8TRt5JHd/fVvwNAKdgjanGZnagDv1tHEPQ1DSTb/Ag==",
            "signing_payload_b64": "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",
            "payload_sha256": "uaDhtI2e7/M5CoNNcthDrIfB1IkMJcolt3e8NxLixKM="
          },
          "tokens_per_usd_hour": 345
        },
        {
          "model": "Sarvam-1",
          "page": "/inferences/sarvam/1",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 62.17,
          "cost_usd_per_million": 3.1624,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "7Mzrnr6rMNlpSm5ORXmlpggnh+MORaGzOqNxSLN1c1pdI/9dI4FTjkkaE3GENmfI16MVzddjWopPHWH0IDYmCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "0mjk6XLJMNmeA5rkOnGSVK8GLYfGbWRqcdgcSRO8xkQ="
          },
          "tokens_per_usd_hour": 88
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_0 (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_0-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_0",
          "engine": "ollama-host",
          "tokens_per_sec": 57.55,
          "cost_usd_per_million": 1.6164,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "X6xSFl9Do3o55VvjwVJO5dOIn2c1W1u3LYxcNXH3CdqPx6/4HXARa+zHBV+Y5qjPXjy0Kk1iwNGSQICwml9oCQ==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NTcuNTUzMjM3MjQwNTk2MTUsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzNjE5MTEwNDQyMTA5MDAwLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwYWQyYmExZmFiYmM4ZTVmMjM4OTJjMDEzNDIyNTc5MGRkMDM1ZDY1MmNlNTQxZGMxNmUwYjA5OGM1NmFmYmI4YSIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoiZGFyd2luL2FybTY0L2FwcGxlLzE2ZyIsIm1vZGVsIjoiaGYuY28vdW5zbG90aC9nZW1tYS00LUUyQi1pdC1HR1VGOlE0XzAiLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6Nywic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjoxNS4xMTY0MzksInRvdGFsX3Rva2VucyI6ODcwfX0=",
            "payload_sha256": "A3KKwoaT+bo1eMQR1EiO0CL8jaH1MA/+Pv6NR8OThUw="
          },
          "tokens_per_usd_hour": 172
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "tokens_per_sec": 53.69,
          "cost_usd_per_million": 3.6987,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "UAbEVKskbftiXCBPpq98EKE67ccZSxX28i8PpJuhscmz+xI/DaBYd5o20oOxPsx3sHE37vDjkB8yZ7/8obRRCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "GuAzHNGJNLDo46XCZWcy7SexOTRTk66Ra3CObYxU5s0="
          },
          "tokens_per_usd_hour": 75
        },
        {
          "model": "Llama-3.2-3B-Instruct-uncensored",
          "page": "/inferences/llama-3-2/3b~chuanli11-llama-3-2-uncensored",
          "build": "IQ3_XS (3-bit) by bartowski",
          "build_page": "/inferences/llama-3-2/3b~chuanli11-llama-3-2-uncensored/bartowski-iq3_xs-ollama",
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-uncensored-GGUF:IQ3_XS",
          "engine": "ollama-host",
          "tokens_per_sec": 46.8,
          "cost_usd_per_million": null,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "b/oo2aQS+/Qgttu+TFayPpm6sBm8CTlGlkjB9xs6p/ss82UbiLV9U5sNGGYoufP8zOjfaQtaNqpoMoUc7YLDDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "Zmy7KqJIynA8GhVSwfxYwPRMiyZwkn1tQ8b1VVzb4g8="
          },
          "tokens_per_usd_hour": null
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "IQ4_XS (4-bit) by bartowski",
          "build_page": "/inferences/phi-4/mini/bartowski-iq4_xs-ollama",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ4_XS",
          "engine": "ollama-host",
          "tokens_per_sec": 36.2,
          "cost_usd_per_million": 5.4463,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "hx1acBZQso+nYFg4CiuuSEh08F2QiviMzROVBjOoL/rl5IdfExrOt1CCHgrgyYQRhtMWwOQHY8IU4zdS6Q0OCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "I4lIVgQ6kUWnZ01OoU16C2ieBRZz7hZkcbWuwyVMtyE="
          },
          "tokens_per_usd_hour": 51
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q4_0 (4-bit) by bartowski",
          "build_page": "/inferences/phi-4/mini/bartowski-q4_0-ollama",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q4_0",
          "engine": "ollama-host",
          "tokens_per_sec": 33.74,
          "cost_usd_per_million": 5.843,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "x7z9ZPKprNiQ3A1wl2wFWNVH9L5yDKBpr73E1l3ThovkhvbrLGND3jWAWd2sXRfWdG2dY5Nmvo9QZX0xQ4iVAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "xR29jlBezGlhrFJPRjm/sFgHjG9fMsg/7UPO0ublYIc="
          },
          "tokens_per_usd_hour": 48
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/phi-4/mini/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 33.7,
          "cost_usd_per_million": 5.8504,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "krq4gBbUnikVohZgyuVVfAabzQ5JC2HIDrElFzyrPLcbexS+jBgrWnBxB+zZVVch0h+kibI3di57mYmMwrEQAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "NiRAaCmNiqGv77eYFDq/H4m9BDBsigLdPw+JhLnSTVo="
          },
          "tokens_per_usd_hour": 47
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q3_K_S (3-bit) by bartowski",
          "build_page": "/inferences/phi-4/mini/bartowski-q3_k_s-ollama",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 28.18,
          "cost_usd_per_million": 3.2992,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "8Xz1GAWAASCaz8+14tIwqc5fiIo7WrZxPRbU6VuPKFVdKDzCEaDKp0AeJ0nE9LYguxZRietvAF3+sG9FEOg2Aw==",
            "signing_payload_b64": "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",
            "payload_sha256": "KHVOPerhOmuf3iqavx9O84HEDTDTO+tHz7sOHBbiJPY="
          },
          "tokens_per_usd_hour": 84
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "IQ2_M (2-bit) by bartowski",
          "build_page": "/inferences/phi-4/mini/bartowski-iq2_m-ollama",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ2_M",
          "engine": "ollama-host",
          "tokens_per_sec": 27.09,
          "cost_usd_per_million": 7.2781,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "5pnmmnCBFX7YVBu9HoguAmWP+UuHO/wHarFN2zDBCHYuXG2Ka10uYsIIcpuHLBpDFwR1hub6M868yNuzENy5Bg==",
            "signing_payload_b64": "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",
            "payload_sha256": "f5aMCnoE37Ik7MwAQNtQ2SvCk0GoQioSuiFw5jINQvA="
          },
          "tokens_per_usd_hour": 38
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "IQ3_XXS (3-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq3_xxs-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "tokens_per_sec": 25.64,
          "cost_usd_per_million": 7.6718,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "IBMT/F7hJO2nqME2KrP/n7yY+nlYuLWtpwc6ibnWXclQaSoL72bjvCVgWe9DcS8bcwOiIowpAdTCxyxLS9/KBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "zw7JoC+W1ts5fP31Khuov5RNyFYYQkMnRNGNrRlI+r0="
          },
          "tokens_per_usd_hour": 36
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "Q5_0 (5-bit) by Qwen",
          "build_page": "/inferences/qwen-3/4b/qwen-q5_0-ollama",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_0",
          "engine": "ollama-host",
          "tokens_per_sec": 22.66,
          "cost_usd_per_million": 8.7188,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "wwU47N3va8t3ehLThYrCmHc2I3QnqTv6T6SYKU9eayS4I7DoN0x81QhpqLcsWzxQFb8z4SBw/6R0UyQ2meApAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "pDfPFm0SPAKBdK+FiIg5yz/KckIG/NEZIFB0815FBNA="
          },
          "tokens_per_usd_hour": 32
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "Q8_0 (8-bit) by lmstudio-community",
          "build_page": "/inferences/phi-4/mini-reasoning/lmstudio-community-q8_0-ollama",
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 22.03,
          "cost_usd_per_million": 8.8724,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "oLcOs1mva6Fmz/UyzJDs93C0Rw0fLUF8MmxM9JejJ5dsPwzr4DxwOuIkJuFh1NpA+RyYeCxXP0bXSyIirZ9EBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "tqbJL0BnbZueOPwGU0LF5uuM6hhffsKnwoKH4BPKSko="
          },
          "tokens_per_usd_hour": 31
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "IQ1_S (1-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq1_s-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ1_S",
          "engine": "ollama-host",
          "tokens_per_sec": 17.18,
          "cost_usd_per_million": 11.4266,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "TgIMAsbsjGLzqcoXlmZgfCLAc6/lk525Hj4VyOI1LwtKD8f67ylHa9TBfFohziMxZGK4oFb70A8iKZoFzt89Aw==",
            "signing_payload_b64": "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",
            "payload_sha256": "38Ohfy5RfNrF/bCYGGoJbq3IwacEhCtW8VQE2rYD/QE="
          },
          "tokens_per_usd_hour": 24
        },
        {
          "model": "Qwen 3 4B",
          "page": "/inferences/qwen-3/4b",
          "build": "IQ1_S (1-bit) by unsloth",
          "build_page": "/inferences/qwen-3/4b/unsloth-iq1_s-ollama",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ1_S",
          "engine": "ollama-host",
          "tokens_per_sec": 14.37,
          "cost_usd_per_million": 13.6614,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "cFgrnzNiTJPDkGcPM0HyrdmSXep0U9UhhFrXfIIj2rM6kCy9u0wvY6w65ME8u78CYZ2KvjslVzZty2HCjHwKBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "f/sMo6oHVgWyeGG0dTH173Ajio62YDIwqV7rfiJjTjY="
          },
          "tokens_per_usd_hour": 20
        },
        {
          "model": "Phi 4 Reasoning-Plus",
          "page": "/inferences/phi-4/reasoning-plus",
          "build": "Q6_K (6-bit) by lmstudio-community",
          "build_page": "/inferences/phi-4/reasoning-plus/lmstudio-community-q6_k-ollama",
          "ref": "hf.co/lmstudio-community/Phi-4-reasoning-plus-GGUF:Q6_K",
          "engine": "ollama-host",
          "tokens_per_sec": 3.26,
          "cost_usd_per_million": 60.0111,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "hFDo9wK6MhBqmtCWuWYsxVFRTfQb0wKoKFQC4brRr2vDevawclDeswNcORKGualHahw4WS7s09YQnv/OoCCQAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "1WnnjF0uxnidjiRXm9j8zaDZQjU/lElGizDI0ATDZ/Q="
          },
          "tokens_per_usd_hour": 5
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "IQ3_XXS (3-bit) by bartowski",
          "build_page": "/inferences/phi-4/mini/bartowski-iq3_xxs-ollama",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "tokens_per_sec": 0.69,
          "cost_usd_per_million": 286.1754,
          "measured_at": "2026-11-02T11:31:50.442Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "93XaUH1gQlcOriBiYNWj/r7IvNDXGUwkPw/kJ8bomBa2dFRrSndPa7bisUbIsE3F74Cj5h2oURZmp885nPqFCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "rzUqPv81DXHp4AxZdpcqdvTB21ICTww0FZBTxfrbJAs="
          },
          "tokens_per_usd_hour": 1
        }
      ],
      "rank": 9,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 100,
        "cpu_all_cores": 93,
        "cpu_one_core": 100,
        "disk_read": 96,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "662a4e",
      "did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
      "name": "Gram ·662a4e",
      "claimed": true,
      "online": false,
      "last_seen": "2026-11-15T07:05:21.952Z",
      "version": "773712f",
      "region": {
        "code": "TW_110",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "AMD EPYC 7B12",
        "cpu_clock_ghz": null,
        "cpus": 4,
        "ram_gb": 31.346603393554688,
        "memory_bandwidth_gb_s": 6.3,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 4982.5,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "amd"
      },
      "sheet": {
        "cpu": {
          "model": "AMD EPYC 7B12",
          "model_raw": "AMD EPYC 7B12",
          "clock_ghz": null,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 512,
          "l3_mb": 16,
          "single_core_hashes_per_sec": 1460.2,
          "multi_core_hashes_per_sec": 3246.1,
          "sustained_ratio": 1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-15T01:38:20.212Z",
          "measured_in": "TW_110",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 6.3,
          "latency_ns": 21.1,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-15T01:38:21.135Z",
          "measured_in": "TW_110",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 4982.5,
          "write_mb_s": 179.3,
          "random_read_iops": 476287,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-11-15T01:38:21.741Z",
          "measured_in": "TW_110",
          "signed": true
        }
      },
      "signed_runs": 35,
      "signed_runs_30d": 17,
      "builds_measured": 35,
      "best_results": 14,
      "first_signed_at": "2026-11-04T03:42:54.890Z",
      "last_signed_at": "2026-11-15T06:44:44.997Z",
      "awake_pct": 100,
      "queue": {
        "measured": 18,
        "failed": 32,
        "queued": 1029
      },
      "download_mib_per_sec": 66.1,
      "uptime_days": [
        {
          "date": "2026-11-04",
          "hours": 21
        },
        {
          "date": "2026-11-05",
          "hours": 17
        },
        {
          "date": "2026-11-06",
          "hours": 18
        },
        {
          "date": "2026-11-07",
          "hours": 20
        },
        {
          "date": "2026-11-08",
          "hours": 24
        },
        {
          "date": "2026-11-09",
          "hours": 24
        },
        {
          "date": "2026-11-10",
          "hours": 23
        },
        {
          "date": "2026-11-11",
          "hours": 24
        },
        {
          "date": "2026-11-12",
          "hours": 24
        },
        {
          "date": "2026-11-13",
          "hours": 9
        },
        {
          "date": "2026-11-14",
          "hours": 17
        },
        {
          "date": "2026-11-15",
          "hours": 8
        }
      ],
      "uptime_pct": 79.5,
      "best_value": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": null,
        "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
        "ref": "llama3.2:1b",
        "engine": "ollama-host",
        "tokens_per_sec": 10.79,
        "cost_usd_per_million": 2.8196,
        "measured_at": "2026-11-14T20:22:46.948Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
          "signature": "YnmWexpPC8JA+V6mE8GoELi444EXbvlieEu8EMkIs3sIi9Ru/YJxvPVDAp2KeeN7xyGH0bGJWR8ATiOhrJ5rBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "9/LgoM2fHYoSl4S+4c/CUnKMNv0+w53a/nJDIz2a6cE="
        },
        "tokens_per_usd_hour": 99
      },
      "opened": [
        {
          "ref": "hf.co/unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_M",
          "family": "qwen-3-5",
          "member": "35b-a3b",
          "at": "2026-11-15T06:44:44.997Z"
        },
        {
          "ref": "hf.co/mradermacher/GLM-5_4layer-i1-GGUF:Q4_K_M",
          "family": "glm-5",
          "member": "4layer~pinaster",
          "at": "2026-11-15T06:44:44.997Z"
        },
        {
          "ref": "hf.co/mradermacher/North-Mini-Code-1.0-i1-GGUF:IQ3_XXS",
          "family": "north-1",
          "member": "mini-code-1-0~mradermacher-i1",
          "at": "2026-11-15T06:44:44.997Z"
        },
        {
          "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:IQ1_M",
          "family": "gpt-oss",
          "member": "20b~davidau-openai-claude-4-5-opus-heretic-uncensored",
          "at": "2026-11-15T06:44:44.997Z"
        },
        {
          "ref": "hf.co/mradermacher/Qwen3.5-27B-kimi-k2.5-Reasoning-Distilled-i1-GGUF:IQ3_XXS",
          "family": "kimi-k2-5",
          "member": "27b-reasoning~mradermacher-qwen3-5-distilled-i1",
          "at": "2026-11-15T06:44:44.997Z"
        },
        {
          "ref": "hf.co/Qwen/Qwen3-VL-2B-Instruct-GGUF:Q4_K_M",
          "family": "qwen-3",
          "member": "vl-2b",
          "at": "2026-11-15T06:44:44.997Z"
        },
        {
          "ref": "hf.co/TeichAI/Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill-GGUF:IQ3_XS",
          "family": "deepseek-v3-2",
          "member": "devstral-small-2505-speciale-distill~teichai",
          "at": "2026-11-15T06:44:44.997Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_S",
          "family": "qwen-3-5",
          "member": "35b-a3b",
          "at": "2026-11-15T06:44:44.997Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3.5-9B-GGUF:Q8_0",
          "family": "qwen-3-5",
          "member": "9b",
          "at": "2026-11-15T06:44:44.997Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-11-04",
          "runs": 442
        },
        {
          "date": "2026-11-05",
          "runs": 365
        },
        {
          "date": "2026-11-06",
          "runs": 117
        },
        {
          "date": "2026-11-07",
          "runs": 692
        },
        {
          "date": "2026-11-08",
          "runs": 742
        },
        {
          "date": "2026-11-09",
          "runs": 924
        },
        {
          "date": "2026-11-10",
          "runs": 799
        },
        {
          "date": "2026-11-11",
          "runs": 969
        },
        {
          "date": "2026-11-12",
          "runs": 510
        },
        {
          "date": "2026-11-13",
          "runs": 306
        },
        {
          "date": "2026-11-14",
          "runs": 561
        },
        {
          "date": "2026-11-15",
          "runs": 34
        }
      ],
      "runs_by_family": [
        {
          "family": "gpt-oss",
          "runs": 7
        },
        {
          "family": "qwen-3-5",
          "runs": 5
        },
        {
          "family": "sd-turbo",
          "runs": 3
        },
        {
          "family": "depth-anything",
          "runs": 2
        },
        {
          "family": "moonshine",
          "runs": 2
        },
        {
          "family": "llama-3-2",
          "runs": 1
        },
        {
          "family": "glm-4-7",
          "runs": 1
        },
        {
          "family": "musicgen",
          "runs": 1
        },
        {
          "family": "sam",
          "runs": 1
        },
        {
          "family": "nllb",
          "runs": 1
        },
        {
          "family": "tiny-sd",
          "runs": 1
        },
        {
          "family": "kokoro",
          "runs": 1
        },
        {
          "family": "whisper",
          "runs": 1
        },
        {
          "family": "wespeaker",
          "runs": 1
        },
        {
          "family": "ecapa-tdnn",
          "runs": 1
        },
        {
          "family": "glm-5",
          "runs": 1
        },
        {
          "family": "north-1",
          "runs": 1
        },
        {
          "family": "kimi-k2-5",
          "runs": 1
        },
        {
          "family": "qwen-3",
          "runs": 1
        },
        {
          "family": "deepseek-v3-2",
          "runs": 1
        },
        {
          "family": "qwen-3-6",
          "runs": 1
        }
      ],
      "rooms": [
        "0oaKlVqk2eNkyh25H78Z",
        "26NRJ3lXsSmqiI4k6N4X",
        "BE_1930",
        "DG1h70GzVONqlxjluvnM",
        "NL_9711",
        "TW_110",
        "US_20511",
        "US_29415",
        "US_51503",
        "US_97058"
      ],
      "awards": [
        {
          "id": "runs-10",
          "label": "10+ signed runs"
        },
        {
          "id": "opener",
          "label": "First to measure 9 builds"
        },
        {
          "id": "best-10",
          "label": "Best result on 14 builds"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": null,
        "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
        "ref": "llama3.2:1b",
        "engine": "ollama-host",
        "tokens_per_sec": 10.79,
        "cost_usd_per_million": 2.8196,
        "measured_at": "2026-11-14T20:22:46.948Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
          "signature": "YnmWexpPC8JA+V6mE8GoELi444EXbvlieEu8EMkIs3sIi9Ru/YJxvPVDAp2KeeN7xyGH0bGJWR8ATiOhrJ5rBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "9/LgoM2fHYoSl4S+4c/CUnKMNv0+w53a/nJDIz2a6cE="
        },
        "tokens_per_usd_hour": 99
      },
      "cheapest": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": null,
        "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
        "ref": "llama3.2:1b",
        "engine": "ollama-host",
        "tokens_per_sec": 10.79,
        "cost_usd_per_million": 2.8196,
        "measured_at": "2026-11-14T20:22:46.948Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
          "signature": "YnmWexpPC8JA+V6mE8GoELi444EXbvlieEu8EMkIs3sIi9Ru/YJxvPVDAp2KeeN7xyGH0bGJWR8ATiOhrJ5rBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "9/LgoM2fHYoSl4S+4c/CUnKMNv0+w53a/nJDIz2a6cE="
        },
        "tokens_per_usd_hour": 99
      },
      "builds": [
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": null,
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "tokens_per_sec": 10.79,
          "cost_usd_per_million": 2.8196,
          "measured_at": "2026-11-14T20:22:46.948Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "YnmWexpPC8JA+V6mE8GoELi444EXbvlieEu8EMkIs3sIi9Ru/YJxvPVDAp2KeeN7xyGH0bGJWR8ATiOhrJ5rBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "9/LgoM2fHYoSl4S+4c/CUnKMNv0+w53a/nJDIz2a6cE="
          },
          "tokens_per_usd_hour": 99
        },
        {
          "model": "Qwen 3 VL-2B",
          "page": "/inferences/qwen-3/vl-2b",
          "build": "Q4_K_M (4-bit) by Qwen",
          "build_page": "/inferences/qwen-3/vl-2b/qwen-q4_k_m-ollama",
          "ref": "hf.co/Qwen/Qwen3-VL-2B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 9.24,
          "cost_usd_per_million": 3.3842,
          "measured_at": "2026-11-15T06:44:44.996Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "SgSuH+BqllkOZEseVft005CqxKyZ6F6Kd1oyNJr+6aSjhisCsw7f8O3bDilr8quGdp+No2dAPXiCKJ6e+em2Bw==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6OS4yNDE3MzE0OTM0NzI5NiwidW5pdCI6InRva2Vucy9zIiwidHMiOjE3OTQ3MjUwODQ5OTY4OTg3NTYsIm5vZGVfZGlkIjoiZGlkOmVwbjowMDI0MDgwMTEyMjA2ZWVmNmI3Mzg0YjY2NDI4MGY3YTYyMmI2YzFhODMxYTQ1MzAyOWQwOTgwOTU2ZWM4MDkwOTU0ZTcyNjYyYTRlIiwiZXh0cmEiOnsiaGFyZHdhcmVfY2xhc3MiOiJsaW51eC9hbWQ2NC9ub25lLzMyZyIsIm1vZGVsIjoiaGYuY28vUXdlbi9Rd2VuMy1WTC0yQi1JbnN0cnVjdC1HR1VGOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50IjoyOSwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjozNjUuNjI0MTI1OTk2OTk5OTYsInRvdGFsX3Rva2VucyI6MzM3OX19",
            "payload_sha256": "nrdAB6QxpwN/9Cla2yZ7FqVDJb4NpYQ6snBietKbRY4="
          },
          "tokens_per_usd_hour": 82
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q4_0 (4-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q4_0-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_0",
          "engine": "ollama-host",
          "tokens_per_sec": 4.56,
          "cost_usd_per_million": 8.6295,
          "measured_at": "2026-11-14T20:22:46.948Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "YyzbZTOjhiM+NIhCD3X67fna6xFNzI6VTHGQZmRY5zqb6y1sSFwAXFlW05zBBwC2U2cAsQy6pb3KGri4hlWVCw==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NC41NTkxNDMyNjk0MjY0NTksInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0Njg3NzY2OTQ4ODUwOTI5LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwZGNkY2NiZDdjYTM0NjFmODUzY2JmNWMwZjc1MjE3ZTVmOTBjOTY4OTY0NTdkOTJiNTliZWQxYjZlZTRlOTRiMiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbm9uZS8zMmciLCJtb2RlbCI6ImhmLmNvL3Vuc2xvdGgvZ3B0LW9zcy0yMGItR0dVRjpRNF8wIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjQyLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjMyMTkuNjgzODYsInRvdGFsX3Rva2VucyI6MTQ2Nzl9fQ==",
            "payload_sha256": "CfyUetf0TNjuJqVUDuD7MgL2vm6jU9oWa4MsYSUBfv4="
          },
          "tokens_per_usd_hour": 32
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 4.41,
          "cost_usd_per_million": 8.9491,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "+G+ybq6tBWwm8EeOJZCgzn79CShJvim8ErkVPpzvObHOUhDPdB/KA3o9E7uGP6DiI32t9OYZET9XE/9EhybnCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "0R0oWGjp4U8rA980GTXB+sz8mG6Wr/4Ig/X4ubJEbIc="
          },
          "tokens_per_usd_hour": 31
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q3_K_S (3-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q3_k_s-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 4.35,
          "cost_usd_per_million": 9.032,
          "measured_at": "2026-11-15T06:44:44.996Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "F0s6Gh5rDuyr4vEDOeR1kqpH1W+yQ1M/KH/J6iKXQs29UGm/mSrfN47JCnAyS5hxolalU0U9RZH9rjzTd/GyBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "x/e4uK04T8R7jWxRm8p5HcufuBBn/lOtxOOeyjHRSI4="
          },
          "tokens_per_usd_hour": 31
        },
        {
          "model": "OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
          "page": "/inferences/gpt-oss/20b~davidau-openai-claude-4-5-opus-heretic-uncensored",
          "build": "IQ1_M (1-bit) by mradermacher",
          "build_page": "/inferences/gpt-oss/20b~davidau-openai-claude-4-5-opus-heretic-uncensored/mradermacher-iq1_m-ollama",
          "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:IQ1_M",
          "engine": "ollama-host",
          "tokens_per_sec": 4.23,
          "cost_usd_per_million": 9.3918,
          "measured_at": "2026-11-15T06:44:44.996Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "PYnjyde4KfUVIYw8HXMH94/Msu3rxXo+eruh51LVfPGig4X+RjfXQH0jb65bpAEnsdI7hszcNSo41TisnozwDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "ywqajNnr4KJBorDevz6q9yIHFs6bw29WBmgtvn00KTk="
          },
          "tokens_per_usd_hour": 30
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q2_K (2-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q2_k-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q2_K",
          "engine": "ollama-host",
          "tokens_per_sec": 4.16,
          "cost_usd_per_million": 9.4507,
          "measured_at": "2026-11-15T06:44:44.996Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "RKQoJp5fVa+d3jp+CBOHgEaNM9UvrXNUuyjlcWyMO+/FeAH2Fgy0mBtD7BbaEm+DjSOG8Wbuqhl62ta43+WJDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "PUqF+vuaVswGiydPi/94HmxVrP1bGe4m3jIO/DhOaNI="
          },
          "tokens_per_usd_hour": 29
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q8_0 (8-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q8_0-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 4.04,
          "cost_usd_per_million": 9.856,
          "measured_at": "2026-10-30T06:35:24.096Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "gi+quH7r/k3h8tXvHURih+O0v8Tf50UIU4u1VlA9rESv5+aX1UyiHHUqhu7c7O0vsxemXusW0clM3VGWg9PbBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "F1WzqmDROeCS7UKDO3BgqNkwhlmvZDMBhI7bj2slhgI="
          },
          "tokens_per_usd_hour": 28
        },
        {
          "model": "Qwen 3.5 35B-A3B",
          "page": "/inferences/qwen-3-5/35b-a3b",
          "build": "Q4_K_S (4-bit) by unsloth",
          "build_page": "/inferences/qwen-3-5/35b-a3b/unsloth-q4_k_s-ollama",
          "ref": "hf.co/unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 3.98,
          "cost_usd_per_million": 11.6834,
          "measured_at": "2026-11-15T06:44:44.996Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "ll8T7vTQKj2IOFF+pgq5X8FSdVYG7S7tEpVjkv5MseljkpvIWxK/HCGbTLUC6suVvFWcfKTe+qjRpxcGWXmmAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "UgzyS5meesDl+TV8ejTVMzsaZH15KN3huzZcsIEWmFc="
          },
          "tokens_per_usd_hour": 24
        },
        {
          "model": "Qwen 3.5 35B-A3B",
          "page": "/inferences/qwen-3-5/35b-a3b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/qwen-3-5/35b-a3b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 3.9,
          "cost_usd_per_million": 12.1846,
          "measured_at": "2026-11-15T06:44:44.996Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "Du0/LK/kmEofPwPX5IBXRRXpL3ctP6QIaploNM7/Wsh1e+7Z5tTLuXPIjY/y2SOSCeor1+glC/2yGRDN0p7uCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "OB3L4nxoTRZGU3Cspqtrp+iqA55rIvR9fygLBvbt0Ts="
          },
          "tokens_per_usd_hour": 23
        },
        {
          "model": "GLM-4.7 Flash",
          "page": "/inferences/glm-4-7/flash",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/glm-4-7/flash/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/GLM-4.7-Flash-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 3.21,
          "cost_usd_per_million": 13.9098,
          "measured_at": "2026-11-15T06:44:44.996Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "osyHIjL41qBWM+G7elAn5fr8RAgFn18d2ZQgtdX08uWiS+YSvNASzqEvHGfZrdVsrV3J+6qUTz496gA71/x/Aw==",
            "signing_payload_b64": "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",
            "payload_sha256": "ZVTi0FWgwZzqp36XBQD39h1uc9u3k3rCtFFanU0NkzI="
          },
          "tokens_per_usd_hour": 20
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q8_K_XL (8-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q8_k_xl-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q8_K_XL",
          "engine": "ollama-host",
          "tokens_per_sec": 3.11,
          "cost_usd_per_million": 13.0639,
          "measured_at": "2026-11-15T06:44:44.996Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "1GSxD6baWLSM851C1oa8ybWrh95aBbcsfdqewrzQ6TTRj3Y2Zn537hHispEQ8kt+PzCS8HxzJHjeB9EhmFe2Cw==",
            "signing_payload_b64": "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",
            "payload_sha256": "U4CHi8mB9sRxoqBbkYJ68ExeDztURvrQxXlk5nVtbe0="
          },
          "tokens_per_usd_hour": 21
        },
        {
          "model": "North 1.0 Mini-Code-1-0",
          "page": "/inferences/north-1/mini-code-1-0",
          "build": "IQ3_XXS (3-bit) by mradermacher",
          "build_page": "/inferences/north-1/mini-code-1-0~mradermacher-i1/mradermacher-iq3_xxs-ollama",
          "ref": "hf.co/mradermacher/North-Mini-Code-1.0-i1-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "tokens_per_sec": 2.15,
          "cost_usd_per_million": 18.4585,
          "measured_at": "2026-11-15T06:44:44.996Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "mvXYbGmkkpE3fb3bJmFTimbVH7pWkgYZ4KmlSNmkPQ1LvbtYcT8uh/KmCpsJxYKPcVHJRGX7wNKZ0gT4j+ppDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "iqJSHOJAuxFLESHFD5j/W7RIfZUN+hkWG4NRGrmbow4="
          },
          "tokens_per_usd_hour": 15
        },
        {
          "model": "Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill",
          "page": "/inferences/deepseek-v3-2/devstral-small-2505-speciale-distill~teichai",
          "build": "IQ3_XS (3-bit) by TeichAI",
          "build_page": "/inferences/deepseek-v3-2/devstral-small-2505-speciale-distill~teichai/teichai-iq3_xs-ollama",
          "ref": "hf.co/TeichAI/Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill-GGUF:IQ3_XS",
          "engine": "ollama-host",
          "tokens_per_sec": 0.36,
          "cost_usd_per_million": 105.2094,
          "measured_at": "2026-11-15T06:44:44.996Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "YKSegtdbk3BFhfs4/FH9yVX6NFoPLjmSlaFS0ZpkFUuDBmD4srmGn7dgiwir/KQtarmq6qTVAiZWCTonIjYlBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "tBhCyLdm8wD9AdCo8KzxsZYvL05cuXS5dlNIpL4gVGI="
          },
          "tokens_per_usd_hour": 3
        }
      ],
      "rank": 10,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 81,
        "cpu_all_cores": 71,
        "cpu_one_core": 82,
        "disk_read": 86,
        "download": 60
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "d68672",
      "did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
      "name": "Gram ·d68672",
      "claimed": true,
      "online": true,
      "last_seen": "2026-11-16T00:43:13.808Z",
      "version": "dbd417d",
      "region": {
        "code": "IN_751030",
        "city": "Bhubaneswar",
        "state": "Odisha"
      },
      "hardware": {
        "os": "darwin",
        "arch": "x86_64",
        "class": "darwin/amd64/apple/8g",
        "cpu_model": "Intel Core i5-5350U",
        "cpu_clock_ghz": 1.8,
        "cpus": 4,
        "ram_gb": 8,
        "memory_bandwidth_gb_s": 0.1,
        "gpu": "apple",
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 89.9,
        "gpu_count": 1,
        "gpu_org": "apple",
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Core i5-5350U",
          "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
          "clock_ghz": 1.8,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 256,
          "l3_mb": 3,
          "single_core_hashes_per_sec": 93.7,
          "multi_core_hashes_per_sec": 195.1,
          "sustained_ratio": 0.7,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-15T23:57:35.534Z",
          "measured_in": "IN_751030",
          "signed": true
        },
        "memory": {
          "total_gb": 8,
          "bandwidth_gb_s": 0.1,
          "latency_ns": 170.8,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-15T23:57:37.012Z",
          "measured_in": "IN_751030",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 89.9,
          "write_mb_s": 44.3,
          "random_read_iops": 25045,
          "fs_type": null,
          "device_model": null,
          "measured_at": "2026-11-15T23:57:40.536Z",
          "measured_in": "IN_751030",
          "signed": true
        }
      },
      "signed_runs": 24,
      "signed_runs_30d": 18,
      "builds_measured": 24,
      "best_results": 10,
      "first_signed_at": "2026-10-28T15:03:55.654Z",
      "last_signed_at": "2026-11-16T00:27:19.107Z",
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 4,
        "queued": 1537
      },
      "download_mib_per_sec": 3.5,
      "uptime_days": [
        {
          "date": "2026-10-30",
          "hours": 2
        },
        {
          "date": "2026-10-31",
          "hours": 17
        },
        {
          "date": "2026-11-01",
          "hours": 3
        },
        {
          "date": "2026-11-05",
          "hours": 17
        },
        {
          "date": "2026-11-06",
          "hours": 24
        },
        {
          "date": "2026-11-07",
          "hours": 24
        },
        {
          "date": "2026-11-08",
          "hours": 9
        },
        {
          "date": "2026-11-09",
          "hours": 5
        },
        {
          "date": "2026-11-10",
          "hours": 24
        },
        {
          "date": "2026-11-11",
          "hours": 24
        },
        {
          "date": "2026-11-12",
          "hours": 24
        },
        {
          "date": "2026-11-13",
          "hours": 17
        },
        {
          "date": "2026-11-14",
          "hours": 24
        },
        {
          "date": "2026-11-15",
          "hours": 24
        },
        {
          "date": "2026-11-16",
          "hours": 1
        }
      ],
      "uptime_pct": 66.4,
      "best_value": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": "Q4_K_M (4-bit) by bartowski",
        "build_page": "/inferences/llama-3-2/1b/bartowski-q4_k_m-ollama",
        "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 4.73,
        "cost_usd_per_million": 6.5625,
        "measured_at": "2026-11-16T00:27:19.106Z",
        "context_tokens": 16384,
        "proof": {
          "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
          "signature": "xeXVtalj4tfLU93gEx2RKmQEB7H0C+MCNagv4wmgc6gaPAft5MrDgo3e1NC44pYDrj/AGC72HJZ574zmotUvBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "7cxiDyJf/0sZM6JbVuSqShplnyJrTQQjL6E2BaTbPfk="
        },
        "tokens_per_usd_hour": 42
      },
      "opened": [
        {
          "ref": "hf.co/openbmb/MiniCPM-o-4_5-gguf:Q5_0",
          "family": "minicpm-o-4-5",
          "member": "base",
          "at": "2026-11-16T00:27:19.107Z"
        },
        {
          "ref": "hf.co/tvall43/Qwen3.6-14B-A3B-FableVibes-GGUF:Q3_K_M",
          "family": "qwen-3-6",
          "member": "14b-a3b~tvall43-fablevibes",
          "at": "2026-11-16T00:27:19.107Z"
        },
        {
          "ref": "qwen2.5:0.5b",
          "family": "qwen-2-5",
          "member": "0.5b",
          "at": "2026-11-16T00:27:19.107Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-28",
          "runs": 14
        },
        {
          "date": "2026-10-30",
          "runs": 42
        },
        {
          "date": "2026-10-31",
          "runs": 350
        },
        {
          "date": "2026-11-05",
          "runs": 208
        },
        {
          "date": "2026-11-06",
          "runs": 320
        },
        {
          "date": "2026-11-07",
          "runs": 1056
        },
        {
          "date": "2026-11-08",
          "runs": 384
        },
        {
          "date": "2026-11-09",
          "runs": 144
        },
        {
          "date": "2026-11-10",
          "runs": 887
        },
        {
          "date": "2026-11-11",
          "runs": 1140
        },
        {
          "date": "2026-11-12",
          "runs": 589
        },
        {
          "date": "2026-11-13",
          "runs": 485
        },
        {
          "date": "2026-11-14",
          "runs": 998
        },
        {
          "date": "2026-11-15",
          "runs": 443
        },
        {
          "date": "2026-11-16",
          "runs": 18
        }
      ],
      "runs_by_family": [
        {
          "family": "phi-4",
          "runs": 10
        },
        {
          "family": "llama-3-2",
          "runs": 4
        },
        {
          "family": "gemma-4",
          "runs": 3
        },
        {
          "family": "other",
          "runs": 3
        },
        {
          "family": "qwen-2-5",
          "runs": 2
        },
        {
          "family": "minicpm-o-4-5",
          "runs": 1
        },
        {
          "family": "qwen-3-6",
          "runs": 1
        }
      ],
      "rooms": [
        "26NRJ3lXsSmqiI4k6N4X",
        "IN_752115"
      ],
      "awards": [
        {
          "id": "runs-10",
          "label": "10+ signed runs"
        },
        {
          "id": "opener",
          "label": "First to measure 3 builds"
        },
        {
          "id": "best-10",
          "label": "Best result on 10 builds"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": "Q4_K_M (4-bit) by bartowski",
        "build_page": "/inferences/llama-3-2/1b/bartowski-q4_k_m-ollama",
        "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 4.73,
        "cost_usd_per_million": 6.5625,
        "measured_at": "2026-11-16T00:27:19.106Z",
        "context_tokens": 16384,
        "proof": {
          "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
          "signature": "xeXVtalj4tfLU93gEx2RKmQEB7H0C+MCNagv4wmgc6gaPAft5MrDgo3e1NC44pYDrj/AGC72HJZ574zmotUvBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "7cxiDyJf/0sZM6JbVuSqShplnyJrTQQjL6E2BaTbPfk="
        },
        "tokens_per_usd_hour": 42
      },
      "cheapest": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": "Q4_K_M (4-bit) by bartowski",
        "build_page": "/inferences/llama-3-2/1b/bartowski-q4_k_m-ollama",
        "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 4.73,
        "cost_usd_per_million": 6.5625,
        "measured_at": "2026-11-16T00:27:19.106Z",
        "context_tokens": 16384,
        "proof": {
          "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
          "signature": "xeXVtalj4tfLU93gEx2RKmQEB7H0C+MCNagv4wmgc6gaPAft5MrDgo3e1NC44pYDrj/AGC72HJZ574zmotUvBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "7cxiDyJf/0sZM6JbVuSqShplnyJrTQQjL6E2BaTbPfk="
        },
        "tokens_per_usd_hour": 42
      },
      "builds": [
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/llama-3-2/1b/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 4.73,
          "cost_usd_per_million": 6.5625,
          "measured_at": "2026-11-16T00:27:19.106Z",
          "context_tokens": 16384,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "xeXVtalj4tfLU93gEx2RKmQEB7H0C+MCNagv4wmgc6gaPAft5MrDgo3e1NC44pYDrj/AGC72HJZ574zmotUvBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "7cxiDyJf/0sZM6JbVuSqShplnyJrTQQjL6E2BaTbPfk="
          },
          "tokens_per_usd_hour": 42
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": null,
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "tokens_per_sec": 3.37,
          "cost_usd_per_million": 39.9708,
          "measured_at": "2026-11-16T00:27:19.106Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "xd8hLSzOrPU/DLkbdsBtG/33kOLVd+irZEX/nDwAgKZcgPdOm4RD8XlZyLXQSZnH6v7rUNnqstW4HoCUu4e8DQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "aRps7eQcPSmCtn4JLzuz5SWTh+NJhBImq4KD5mhG5nw="
          },
          "tokens_per_usd_hour": 7
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": "IQ1_S (1-bit) by maziyarpanahi",
          "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-iq1_s-ollama",
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:IQ1_S",
          "engine": "ollama-host",
          "tokens_per_sec": 3.26,
          "cost_usd_per_million": 9.3281,
          "measured_at": "2026-11-16T00:27:19.106Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "SpW5/M0bkEYuxG2s6CD/eUez5QpwFbFF8abJOI0SF2/6IVIyygTNqXo3Tf6PdCnAJdN7xNS+4i6M/bbliZekCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "qP0yV3jyFo3b5KApQMJNiR7uuvjnVSOlK19XpzQJGHA="
          },
          "tokens_per_usd_hour": 30
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q3_K_M (3-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q3_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q3_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 2.78,
          "cost_usd_per_million": 11.5908,
          "measured_at": "2026-11-16T00:27:19.106Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "PQIevpHHIJq7FIIU/zB39melUPlvBe6fyOs5bNdTiquVHiYQx5m6tudJxXrsqtgFW1+RL6qDCPX30rY+E9rRAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "BC/6GOLalZ0BS2F3c5ndM0SLFG03Umqq547zJo6sgns="
          },
          "tokens_per_usd_hour": 24
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q3_K_L (3-bit) by maziyarpanahi",
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q3_k_l-ollama",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "tokens_per_sec": 2.2,
          "cost_usd_per_million": 13.851,
          "measured_at": "2026-11-16T00:27:19.106Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "mu9Xqlwj5QnUrGKvvBo4uSzWdN0mApaH848gtB9ZuhkOJybjp1OOg5ukwpKxns+TNs4N06CGGDRgwWPRbUIwCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "Dl9vP8+iQMF4OzZquCYd6meUIJ97lsBlH4HTtgNH3/4="
          },
          "tokens_per_usd_hour": 20
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 2.18,
          "cost_usd_per_million": 14.8637,
          "measured_at": "2026-11-16T00:27:19.106Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "SD/2MwO6SAmRC9FzH+Yex+vRUEjvBzKtDzcmztrdEWXxI12t1DXhhwnGUnSPM+dcgXXOC6v8z3qgzJpY0b4JAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "vqOr+XhSD4QjGsTzXpbR3XwvKJNhmWa5uj2/g0QG8rs="
          },
          "tokens_per_usd_hour": 19
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "Q3_K_S (3-bit) by maziyarpanahi",
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q3_k_s-ollama",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 2.17,
          "cost_usd_per_million": 14.0402,
          "measured_at": "2026-11-16T00:27:19.106Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "im4DmT2oyHpzCWSA/HqOmUpmlDKFWe802H3/bs2cHY/g4kvzEXDbjeL06f+Xk7dK6J20wYgSFLHq5R7KhW28Dg==",
            "signing_payload_b64": "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",
            "payload_sha256": "e9ig+TjBpsOPX+7wK1J58zCA3krWI8QZfGOBd3zz29c="
          },
          "tokens_per_usd_hour": 20
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 1.85,
          "cost_usd_per_million": 17.4407,
          "measured_at": "2026-11-16T00:27:19.106Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "tqZ75GjCOqnMPnoK21BpdHREBPxjFVQk64Hiwr5hdjyRsDYon+Wo+0UoXxKRWAsPVibx9SDEybqkagrb8BQyBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "kvdmc0V7b5auudEbUfuRL/S2TQKce7MmvP3YaEvzzGo="
          },
          "tokens_per_usd_hour": 16
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": "IQ3_XS (3-bit) by maziyarpanahi",
          "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-iq3_xs-ollama",
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:IQ3_XS",
          "engine": "ollama-host",
          "tokens_per_sec": 1.48,
          "cost_usd_per_million": 20.6089,
          "measured_at": "2026-11-16T00:27:19.106Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "JfIHWmj1EH4cJR9v4NAu/XbQRQSJKHXHnPKxT31jtg00FHAT+3ryZu/4Qonsex0udIs5ksxpg9gcpwwbVrZPDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "v7HypU3osEpbQ7DnJilzW9FaIvs16oSZ859bh3Ybhyg="
          },
          "tokens_per_usd_hour": 14
        },
        {
          "model": "Phi 4 Mini",
          "page": "/inferences/phi-4/mini",
          "build": "IQ2_M (2-bit) by bartowski",
          "build_page": "/inferences/phi-4/mini/bartowski-iq2_m-ollama",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ2_M",
          "engine": "ollama-host",
          "tokens_per_sec": 1.06,
          "cost_usd_per_million": 30.3882,
          "measured_at": "2026-11-16T00:27:19.106Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "5vwtxvAzTU4q5XVNX6H9JR4bfv+L6Ltbtbmf/8BMHCBCpe/OPfE4rBA5roxUR10SjGHCT/O9Y2PJw0aqbqgVCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "yAC5dhK6YI+EbcYMwzCYGrf5HT3kWlCcbY9GGkWVjM8="
          },
          "tokens_per_usd_hour": 9
        }
      ],
      "rank": 11,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 0,
        "cpu_all_cores": 14,
        "cpu_one_core": 11,
        "disk_read": 4,
        "download": 0
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "d061c9",
      "did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
      "name": "Gram ·d061c9",
      "claimed": false,
      "online": false,
      "last_seen": "2026-10-29T09:02:07.675Z",
      "version": "43f84dc",
      "region": {
        "code": "IN_560001",
        "city": "Bengaluru",
        "state": "Karnataka"
      },
      "hardware": {
        "os": "windows",
        "arch": "x86_64",
        "class": "windows/amd64/amd/8g",
        "cpu_model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
        "cpu_clock_ghz": null,
        "cpus": 8,
        "ram_gb": 6.950382232666016,
        "memory_bandwidth_gb_s": 6,
        "gpu": "AMD Radeon(TM) Vega 8 Mobile Graphics",
        "vram_gb": 1,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 2347.3,
        "gpu_count": 1,
        "gpu_org": "amd",
        "cpu_org": "amd"
      },
      "sheet": {
        "cpu": {
          "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
          "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
          "clock_ghz": null,
          "cores_logical": 8,
          "cores_physical": null,
          "threads_per_core": null,
          "sockets": null,
          "l2_kb": null,
          "l3_mb": null,
          "single_core_hashes_per_sec": 997.6,
          "multi_core_hashes_per_sec": 8516.9,
          "sustained_ratio": 1.2,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-10-29T08:14:32.559Z",
          "measured_in": "IN_560001",
          "signed": true
        },
        "memory": {
          "total_gb": 7,
          "bandwidth_gb_s": 6,
          "latency_ns": 61.5,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-10-29T08:14:33.251Z",
          "measured_in": "IN_560001",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 2347.3,
          "write_mb_s": 204.2,
          "random_read_iops": 96070,
          "fs_type": null,
          "device_model": null,
          "measured_at": "2026-10-29T08:14:33.869Z",
          "measured_in": "IN_560001",
          "signed": true
        }
      },
      "signed_runs": 17,
      "signed_runs_30d": 9,
      "builds_measured": 17,
      "best_results": 5,
      "first_signed_at": "2026-10-28T15:49:11.566Z",
      "last_signed_at": "2026-10-29T09:04:03.140Z",
      "awake_pct": 100,
      "queue": {
        "measured": 4,
        "failed": 9,
        "queued": 745
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-29",
          "hours": 2
        }
      ],
      "uptime_pct": 8.3,
      "best_value": {
        "model": "Gemma 3 270M",
        "page": "/inferences/gemma-3/270m",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 27.34,
        "cost_usd_per_million": 6.412,
        "measured_at": "2026-10-29T09:04:03.140Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
          "signature": "GLnSBD5ynTuGblHtKb33m4ubr9VqLhS0BBv1ntoUsasiXw0hjsgDlonfSlamGyX7e06BtkJE3TXmajlN/h8vBw==",
          "signing_payload_b64": "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",
          "payload_sha256": "i7cif1bMTlpocFTMyueR/+vOcEjh9BAhv1dDjBazT7A="
        },
        "tokens_per_usd_hour": 43
      },
      "opened": [
        {
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q4_K_M",
          "family": "phi-4",
          "member": "mini",
          "at": "2026-10-29T09:04:03.140Z"
        },
        {
          "ref": "hf.co/MaziyarPanahi/Qwen3-0.6B-GGUF:Q2_K",
          "family": "qwen-3",
          "member": "0.6b",
          "at": "2026-10-29T09:04:03.140Z"
        },
        {
          "ref": "hf.co/mistralai/Ministral-3-3B-Instruct-2512-GGUF:Q4_K_M",
          "family": "ministral-3",
          "member": "3b-instruct-2512",
          "at": "2026-10-29T09:04:03.140Z"
        },
        {
          "ref": "hf.co/openbmb/MiniCPM-V-4-gguf:Q4_K_M",
          "family": "minicpm-v-4",
          "member": "base",
          "at": "2026-10-29T09:04:03.140Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-28",
          "runs": 9
        },
        {
          "date": "2026-10-29",
          "runs": 36
        }
      ],
      "runs_by_family": [
        {
          "family": "phi-4",
          "runs": 5
        },
        {
          "family": "depth-anything",
          "runs": 2
        },
        {
          "family": "llama-3-2",
          "runs": 1
        },
        {
          "family": "musicgen",
          "runs": 1
        },
        {
          "family": "sam",
          "runs": 1
        },
        {
          "family": "sarvam",
          "runs": 1
        },
        {
          "family": "gemma-3",
          "runs": 1
        },
        {
          "family": "mms-tts",
          "runs": 1
        },
        {
          "family": "embeddinggemma",
          "runs": 1
        },
        {
          "family": "qwen-3",
          "runs": 1
        },
        {
          "family": "ministral-3",
          "runs": 1
        },
        {
          "family": "minicpm-v-4",
          "runs": 1
        }
      ],
      "rooms": [
        "IN_560001",
        "IN_562130"
      ],
      "awards": [
        {
          "id": "runs-10",
          "label": "10+ signed runs"
        },
        {
          "id": "opener",
          "label": "First to measure 4 builds"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "Gemma 3 270M",
        "page": "/inferences/gemma-3/270m",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 27.34,
        "cost_usd_per_million": 6.412,
        "measured_at": "2026-10-29T09:04:03.140Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
          "signature": "GLnSBD5ynTuGblHtKb33m4ubr9VqLhS0BBv1ntoUsasiXw0hjsgDlonfSlamGyX7e06BtkJE3TXmajlN/h8vBw==",
          "signing_payload_b64": "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",
          "payload_sha256": "i7cif1bMTlpocFTMyueR/+vOcEjh9BAhv1dDjBazT7A="
        },
        "tokens_per_usd_hour": 43
      },
      "cheapest": {
        "model": "Gemma 3 270M",
        "page": "/inferences/gemma-3/270m",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 27.34,
        "cost_usd_per_million": 6.412,
        "measured_at": "2026-10-29T09:04:03.140Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
          "signature": "GLnSBD5ynTuGblHtKb33m4ubr9VqLhS0BBv1ntoUsasiXw0hjsgDlonfSlamGyX7e06BtkJE3TXmajlN/h8vBw==",
          "signing_payload_b64": "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",
          "payload_sha256": "i7cif1bMTlpocFTMyueR/+vOcEjh9BAhv1dDjBazT7A="
        },
        "tokens_per_usd_hour": 43
      },
      "builds": [
        {
          "model": "Gemma 3 270M",
          "page": "/inferences/gemma-3/270m",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 27.34,
          "cost_usd_per_million": 6.412,
          "measured_at": "2026-10-29T09:04:03.140Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
            "signature": "GLnSBD5ynTuGblHtKb33m4ubr9VqLhS0BBv1ntoUsasiXw0hjsgDlonfSlamGyX7e06BtkJE3TXmajlN/h8vBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "i7cif1bMTlpocFTMyueR/+vOcEjh9BAhv1dDjBazT7A="
          },
          "tokens_per_usd_hour": 43
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": null,
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "tokens_per_sec": 5.62,
          "cost_usd_per_million": 31.1559,
          "measured_at": "2026-10-29T09:04:03.140Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
            "signature": "2Q//JA9gQANmknVAW7mbRxfmRtQbcA2w0mdGysoT3hwNn/ViPWMv1vhtQFw+RjkgZbYUF7vIjOn1fWsujlBqDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "F1N2EFfzZqANMjAs9vMbJvh7roRWtG0s4rOX+imwhCI="
          },
          "tokens_per_usd_hour": 9
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "Q4_K_M (4-bit) by lmstudio-community",
          "build_page": "/inferences/phi-4/mini-reasoning/lmstudio-community-q4_k_m-ollama",
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 4.75,
          "cost_usd_per_million": 36.8345,
          "measured_at": "2026-10-29T09:04:03.140Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
            "signature": "bz55s8iLtRwoJxOB9SdY7afyef7DIgX/JsCEU1fG2S1Ql+H6Knbf79FmxURFX8A1kebBXMRW5wZQURTa5etqDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "5GSE+TKMXES9NNa4WCvTNhc6p7CNe942CMIV3ETrgCk="
          },
          "tokens_per_usd_hour": 8
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "Q3_K_L (3-bit) by lmstudio-community",
          "build_page": "/inferences/phi-4/mini-reasoning/lmstudio-community-q3_k_l-ollama",
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "tokens_per_sec": 3.85,
          "cost_usd_per_million": 45.5148,
          "measured_at": "2026-10-29T09:04:03.140Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
            "signature": "HS/TRAZZPq/fkBoHbvUMR9LshMP4xyK3ljVAKV9RW9BWk/XkMlDOYLgUCxTFMI/WoRe19yY8lFDp49J48dN+Dg==",
            "signing_payload_b64": "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",
            "payload_sha256": "pVDxiu9t7Zf3V7Fn4FnHMAn0EixHjlaeEcGoLzk3+Ng="
          },
          "tokens_per_usd_hour": 6
        },
        {
          "model": "MiniCPM-V 4 Base",
          "page": "/inferences/minicpm-v-4/base",
          "build": "Q4_K_M (4-bit) by openbmb",
          "build_page": "/inferences/minicpm-v-4/base/openbmb-q4_k_m-ollama",
          "ref": "hf.co/openbmb/MiniCPM-V-4-gguf:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 3.73,
          "cost_usd_per_million": 47.3964,
          "measured_at": "2026-10-29T09:04:03.140Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
            "signature": "XNSd6EbZlPSEliFS/q7+ArudNTCRwnpJidH8e0SNBk+HWVAJosM9Rd0X3Jjn/TKqviPx/JhrgkEJAhJYo2kbDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "/idEGoz2O6T6vdumwKuUNxUKFZqdTANGKDjFLz3SFto="
          },
          "tokens_per_usd_hour": 6
        }
      ],
      "rank": 12,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 78,
        "cpu_all_cores": 86,
        "cpu_one_core": 75,
        "disk_read": 46,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "4e94b2",
      "did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
      "name": "Gram ·4e94b2",
      "claimed": true,
      "online": false,
      "last_seen": "2026-11-14T20:35:54.507Z",
      "version": "773712f",
      "region": {
        "code": "US_97058",
        "city": "The Dalles",
        "state": "Oregon"
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2.2,
        "cpus": 4,
        "ram_gb": 31.346603393554688,
        "memory_bandwidth_gb_s": 1.7,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 2202.4,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
          "clock_ghz": 2.2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 256,
          "l3_mb": 55,
          "single_core_hashes_per_sec": 185.9,
          "multi_core_hashes_per_sec": 588,
          "sustained_ratio": 1.4,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-14T19:37:12.663Z",
          "measured_in": "US_97058",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 1.7,
          "latency_ns": 71.3,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-14T19:37:15.832Z",
          "measured_in": "US_97058",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 2202.4,
          "write_mb_s": 15.5,
          "random_read_iops": 196685,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-11-14T19:37:24.078Z",
          "measured_in": "US_97058",
          "signed": true
        }
      },
      "signed_runs": 11,
      "signed_runs_30d": 8,
      "builds_measured": 11,
      "best_results": 8,
      "first_signed_at": "2026-11-05T06:47:05.641Z",
      "last_signed_at": "2026-11-14T20:22:46.949Z",
      "awake_pct": 100,
      "queue": {
        "measured": 1,
        "failed": 7,
        "queued": 397
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-11-05",
          "hours": 18
        },
        {
          "date": "2026-11-06",
          "hours": 24
        },
        {
          "date": "2026-11-07",
          "hours": 14
        },
        {
          "date": "2026-11-08",
          "hours": 24
        },
        {
          "date": "2026-11-09",
          "hours": 24
        },
        {
          "date": "2026-11-10",
          "hours": 13
        },
        {
          "date": "2026-11-11",
          "hours": 24
        },
        {
          "date": "2026-11-12",
          "hours": 24
        },
        {
          "date": "2026-11-13",
          "hours": 16
        },
        {
          "date": "2026-11-14",
          "hours": 14
        }
      ],
      "uptime_pct": 81.3,
      "best_value": {
        "model": "MiniCPM-V 4.7 1B",
        "page": "/inferences/minicpm-v-4-7/1b",
        "build": "Q8_0 (8-bit) by openbmb",
        "build_page": "/inferences/minicpm-v-4-7/1b/openbmb-q8_0-ollama",
        "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q8_0",
        "engine": "ollama-host",
        "tokens_per_sec": 14.97,
        "cost_usd_per_million": 2.0879,
        "measured_at": "2026-11-14T20:22:46.948Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
          "signature": "DDoNBNp5bGsCF3kCmz17IPxxsHpmPfkPrgAtP/ADhLj4mCJ3PexOIhPVGr4HKAl6UzfYDivFdH3eO/DCZBeSCA==",
          "signing_payload_b64": "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",
          "payload_sha256": "3JxU2y7bmdoj2i4Y1ZQOpAqkqY0GqMgjh+k8ZVXqeu8="
        },
        "tokens_per_usd_hour": 133
      },
      "opened": [
        {
          "ref": "hf.co/huihui-ai/Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF:Q2_K",
          "family": "qwen-3-6",
          "member": "35b-a3b~huihui-ai-abliterated-mtp",
          "at": "2026-11-14T20:22:46.949Z"
        },
        {
          "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q3_K_M",
          "family": "qwen-3-6",
          "member": "35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved",
          "at": "2026-11-14T20:22:46.949Z"
        },
        {
          "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q4_K_M",
          "family": "qwen-3-6",
          "member": "35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved",
          "at": "2026-11-14T20:22:46.949Z"
        },
        {
          "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q4_K_S",
          "family": "qwen-3-6",
          "member": "35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved",
          "at": "2026-11-14T20:22:46.949Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-11-05",
          "runs": 64
        },
        {
          "date": "2026-11-06",
          "runs": 133
        },
        {
          "date": "2026-11-07",
          "runs": 238
        },
        {
          "date": "2026-11-08",
          "runs": 375
        },
        {
          "date": "2026-11-09",
          "runs": 400
        },
        {
          "date": "2026-11-10",
          "runs": 192
        },
        {
          "date": "2026-11-11",
          "runs": 440
        },
        {
          "date": "2026-11-12",
          "runs": 240
        },
        {
          "date": "2026-11-13",
          "runs": 272
        },
        {
          "date": "2026-11-14",
          "runs": 192
        }
      ],
      "runs_by_family": [
        {
          "family": "qwen-3-6",
          "runs": 5
        },
        {
          "family": "llama-3-2",
          "runs": 1
        },
        {
          "family": "nllb",
          "runs": 1
        },
        {
          "family": "sd-turbo",
          "runs": 1
        },
        {
          "family": "whisper",
          "runs": 1
        },
        {
          "family": "gpt-oss",
          "runs": 1
        },
        {
          "family": "minicpm-v-4-7",
          "runs": 1
        }
      ],
      "rooms": [
        "asPepgKbRT1dsKNkRGZd",
        "NL_9711",
        "US_20511",
        "US_29415",
        "US_51503",
        "US_97058"
      ],
      "awards": [
        {
          "id": "runs-10",
          "label": "10+ signed runs"
        },
        {
          "id": "opener",
          "label": "First to measure 4 builds"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "MiniCPM-V 4.7 1B",
        "page": "/inferences/minicpm-v-4-7/1b",
        "build": "Q8_0 (8-bit) by openbmb",
        "build_page": "/inferences/minicpm-v-4-7/1b/openbmb-q8_0-ollama",
        "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q8_0",
        "engine": "ollama-host",
        "tokens_per_sec": 14.97,
        "cost_usd_per_million": 2.0879,
        "measured_at": "2026-11-14T20:22:46.948Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
          "signature": "DDoNBNp5bGsCF3kCmz17IPxxsHpmPfkPrgAtP/ADhLj4mCJ3PexOIhPVGr4HKAl6UzfYDivFdH3eO/DCZBeSCA==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTQuOTY5ODY4MDc2NjI5MDY0LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5NDY4Nzc2Njk0ODg1MDkyOSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMGRjZGNjYmQ3Y2EzNDYxZjg1M2NiZjVjMGY3NTIxN2U1ZjkwYzk2ODk2NDU3ZDkyYjU5YmVkMWI2ZWU0ZTk0YjIiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L25vbmUvMzJnIiwibW9kZWwiOiJoZi5jby9vcGVuYm1iL01pbmlDUE0tVi00LjctMUItZ2d1ZjpROF8wIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjYwLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjIyMC4xNzU2MjA5OTYsInRvdGFsX3Rva2VucyI6MzI5Nn19",
          "payload_sha256": "3JxU2y7bmdoj2i4Y1ZQOpAqkqY0GqMgjh+k8ZVXqeu8="
        },
        "tokens_per_usd_hour": 133
      },
      "cheapest": {
        "model": "MiniCPM-V 4.7 1B",
        "page": "/inferences/minicpm-v-4-7/1b",
        "build": "Q8_0 (8-bit) by openbmb",
        "build_page": "/inferences/minicpm-v-4-7/1b/openbmb-q8_0-ollama",
        "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q8_0",
        "engine": "ollama-host",
        "tokens_per_sec": 14.97,
        "cost_usd_per_million": 2.0879,
        "measured_at": "2026-11-14T20:22:46.948Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
          "signature": "DDoNBNp5bGsCF3kCmz17IPxxsHpmPfkPrgAtP/ADhLj4mCJ3PexOIhPVGr4HKAl6UzfYDivFdH3eO/DCZBeSCA==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTQuOTY5ODY4MDc2NjI5MDY0LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5NDY4Nzc2Njk0ODg1MDkyOSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMGRjZGNjYmQ3Y2EzNDYxZjg1M2NiZjVjMGY3NTIxN2U1ZjkwYzk2ODk2NDU3ZDkyYjU5YmVkMWI2ZWU0ZTk0YjIiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L25vbmUvMzJnIiwibW9kZWwiOiJoZi5jby9vcGVuYm1iL01pbmlDUE0tVi00LjctMUItZ2d1ZjpROF8wIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjYwLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjIyMC4xNzU2MjA5OTYsInRvdGFsX3Rva2VucyI6MzI5Nn19",
          "payload_sha256": "3JxU2y7bmdoj2i4Y1ZQOpAqkqY0GqMgjh+k8ZVXqeu8="
        },
        "tokens_per_usd_hour": 133
      },
      "builds": [
        {
          "model": "MiniCPM-V 4.7 1B",
          "page": "/inferences/minicpm-v-4-7/1b",
          "build": "Q8_0 (8-bit) by openbmb",
          "build_page": "/inferences/minicpm-v-4-7/1b/openbmb-q8_0-ollama",
          "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q8_0",
          "engine": "ollama-host",
          "tokens_per_sec": 14.97,
          "cost_usd_per_million": 2.0879,
          "measured_at": "2026-11-14T20:22:46.948Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "DDoNBNp5bGsCF3kCmz17IPxxsHpmPfkPrgAtP/ADhLj4mCJ3PexOIhPVGr4HKAl6UzfYDivFdH3eO/DCZBeSCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "3JxU2y7bmdoj2i4Y1ZQOpAqkqY0GqMgjh+k8ZVXqeu8="
          },
          "tokens_per_usd_hour": 133
        },
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": null,
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "tokens_per_sec": 10.79,
          "cost_usd_per_million": 2.8196,
          "measured_at": "2026-11-14T20:22:46.948Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "YnmWexpPC8JA+V6mE8GoELi444EXbvlieEu8EMkIs3sIi9Ru/YJxvPVDAp2KeeN7xyGH0bGJWR8ATiOhrJ5rBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "9/LgoM2fHYoSl4S+4c/CUnKMNv0+w53a/nJDIz2a6cE="
          },
          "tokens_per_usd_hour": 99
        },
        {
          "model": "Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF",
          "page": "/inferences/qwen-3-6/35b-a3b~huihui-ai-abliterated-mtp",
          "build": "Q2_K (2-bit) by huihui-ai",
          "build_page": "/inferences/qwen-3-6/35b-a3b~huihui-ai-abliterated-mtp/huihui-ai-q2_k-ollama",
          "ref": "hf.co/huihui-ai/Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF:Q2_K",
          "engine": "ollama-host",
          "tokens_per_sec": 5,
          "cost_usd_per_million": 8.1457,
          "measured_at": "2026-11-14T20:22:46.948Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "Qx6tfy6v914JGxzCFocDcsVZ5t/8lZ6vavl+giiW31Mq+vVESnSznpVelFwDxUXg2FW/vxg7uGPTeyoVdGntBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "j3kCJ/BpEf4p9Wx3upEBKBigpvY0nO8INwQlVebUDeU="
          },
          "tokens_per_usd_hour": 34
        },
        {
          "model": "Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
          "page": "/inferences/qwen-3-6/35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved",
          "build": "Q4_K_M (4-bit) by llmfan46",
          "build_page": "/inferences/qwen-3-6/35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved/llmfan46-q4_k_m-ollama",
          "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 4.85,
          "cost_usd_per_million": 9.7638,
          "measured_at": "2026-11-14T20:22:46.948Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "WN4iNJwRNUhu2+5s3dXJh22mY+r6fzai73ExjbgOV6czlBNQot7uqwyDDHAMHruHbcUoH2f6YN/lNmcDYkn7Cg==",
            "signing_payload_b64": "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",
            "payload_sha256": "lfXVJz9fTBsyK2mY2vgif704gHEGO90xq6bwKgAJ49c="
          },
          "tokens_per_usd_hour": 28
        },
        {
          "model": "Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
          "page": "/inferences/qwen-3-6/35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved",
          "build": "Q3_K_M (3-bit) by llmfan46",
          "build_page": "/inferences/qwen-3-6/35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved/llmfan46-q3_k_m-ollama",
          "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q3_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 4.83,
          "cost_usd_per_million": 9.0744,
          "measured_at": "2026-11-14T20:22:46.948Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "Y1NbHgJQ/4PEDQeSrJuccaqwUYeKO/oPhq47dAyzBA2yJ32EoWadilE4V/BlnMetdw3Z9J4wV4sKhHDNV9erBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "RcN3JWbOSr5kRjFeICrWyn+O70FtCptdmZ8MjqsK3JE="
          },
          "tokens_per_usd_hour": 31
        },
        {
          "model": "Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
          "page": "/inferences/qwen-3-6/35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved",
          "build": "Q4_K_S (4-bit) by llmfan46",
          "build_page": "/inferences/qwen-3-6/35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved/llmfan46-q4_k_s-ollama",
          "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q4_K_S",
          "engine": "ollama-host",
          "tokens_per_sec": 4.81,
          "cost_usd_per_million": 9.6125,
          "measured_at": "2026-11-14T20:22:46.948Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "1lVy2ogFSB1naP+mYMnc1y3T2J+Q3zyHtCO6+6FdAZ+L6t+kz7DcP9KeTUNFdY8ct2agePktB2B8jj9G/jCQAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "iCxNCfxqp9SUaRneojSyd6Te9RqIVXEBXHHsxabP7vY="
          },
          "tokens_per_usd_hour": 29
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q4_0 (4-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q4_0-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_0",
          "engine": "ollama-host",
          "tokens_per_sec": 4.56,
          "cost_usd_per_million": 8.6295,
          "measured_at": "2026-11-14T20:22:46.948Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "YyzbZTOjhiM+NIhCD3X67fna6xFNzI6VTHGQZmRY5zqb6y1sSFwAXFlW05zBBwC2U2cAsQy6pb3KGri4hlWVCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "CfyUetf0TNjuJqVUDuD7MgL2vm6jU9oWa4MsYSUBfv4="
          },
          "tokens_per_usd_hour": 32
        },
        {
          "model": "Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF",
          "page": "/inferences/qwen-3-6/35b-a3b~huihui-ai-abliterated-mtp",
          "build": "Q6_K (6-bit) by huihui-ai",
          "build_page": "/inferences/qwen-3-6/35b-a3b~huihui-ai-abliterated-mtp/huihui-ai-q6_k-ollama",
          "ref": "hf.co/huihui-ai/Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF:Q6_K",
          "engine": "ollama-host",
          "tokens_per_sec": 4.29,
          "cost_usd_per_million": 12.3819,
          "measured_at": "2026-11-14T20:22:46.948Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "xt9ES5v2myeNju0C0/oQI786YQUEMxKxNpPpbPmWpzzTRJnOHbS7qby1zi/guxzZNOh+j2ad+ED3z+Y8pOVPCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "A7asW5Anl42qeNaOTyfVt5iKgX98VQhbPu/Ahb7aZ+k="
          },
          "tokens_per_usd_hour": 22
        }
      ],
      "rank": 13,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 37,
        "cpu_all_cores": 54,
        "cpu_one_core": 43,
        "disk_read": 36,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "0548a8",
      "did": "did:epn:002408011220ab92819c6391e87d553e27cec9a18a52937a84a123bf752d00c2aa2bee0548a8",
      "name": "Gram ·0548a8",
      "claimed": false,
      "online": false,
      "last_seen": "2026-10-21T18:54:26.305Z",
      "version": "7b3b602",
      "region": null,
      "hardware": {
        "os": "darwin",
        "arch": "arm64",
        "class": "darwin/arm64/apple/8g",
        "cpu_model": null,
        "cpu_clock_ghz": null,
        "cpus": 8,
        "ram_gb": 8,
        "memory_bandwidth_gb_s": null,
        "gpu": "Apple M1",
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": null,
        "gpu_count": 1,
        "gpu_org": "apple",
        "cpu_org": null
      },
      "sheet": null,
      "signed_runs": 9,
      "signed_runs_30d": 0,
      "builds_measured": 9,
      "best_results": 3,
      "first_signed_at": "2026-10-08T15:04:04.376Z",
      "last_signed_at": "2026-10-08T15:04:04.376Z",
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 7,
        "queued": 144
      },
      "download_mib_per_sec": null,
      "uptime_days": [],
      "uptime_pct": null,
      "best_value": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": null,
        "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
        "ref": "llama3.2:1b",
        "engine": "ollama-host",
        "tokens_per_sec": 20.37,
        "cost_usd_per_million": 8.5986,
        "measured_at": "2026-10-08T15:04:04.375Z",
        "context_tokens": 32768,
        "proof": {
          "node_did": "did:epn:002408011220ab92819c6391e87d553e27cec9a18a52937a84a123bf752d00c2aa2bee0548a8",
          "signature": "eebpktI5uYJizQhh7Sn0Iv+0jvsz9CWBbeHIi9IeoqUpY6GQXP6bH4ZrCTw4e9xfcfjk3Uz2q9+r0kPMHICLAQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "qVEsr7bcara1XYeXsf0n2J0VT2RIdNte7F5A6GDGXWE="
        },
        "tokens_per_usd_hour": 32
      },
      "opened": [
        {
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ1_S",
          "family": "phi-4",
          "member": "mini-reasoning",
          "at": "2026-10-08T15:04:04.376Z"
        },
        {
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
          "family": "phi-4",
          "member": "mini-reasoning",
          "at": "2026-10-08T15:04:04.376Z"
        },
        {
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q3_K_L",
          "family": "phi-4",
          "member": "mini-reasoning",
          "at": "2026-10-08T15:04:04.376Z"
        },
        {
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q4_K_M",
          "family": "phi-4",
          "member": "mini-reasoning",
          "at": "2026-10-08T15:04:04.376Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-08",
          "runs": 9
        }
      ],
      "runs_by_family": [
        {
          "family": "phi-4",
          "runs": 5
        },
        {
          "family": "qwen-3",
          "runs": 2
        },
        {
          "family": "llama-3-2",
          "runs": 1
        },
        {
          "family": "gemma-4",
          "runs": 1
        }
      ],
      "rooms": [
        "IN_500001"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "opener",
          "label": "First to measure 4 builds"
        }
      ],
      "fastest": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": null,
        "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
        "ref": "llama3.2:1b",
        "engine": "ollama-host",
        "tokens_per_sec": 20.37,
        "cost_usd_per_million": 8.5986,
        "measured_at": "2026-10-08T15:04:04.375Z",
        "context_tokens": 32768,
        "proof": {
          "node_did": "did:epn:002408011220ab92819c6391e87d553e27cec9a18a52937a84a123bf752d00c2aa2bee0548a8",
          "signature": "eebpktI5uYJizQhh7Sn0Iv+0jvsz9CWBbeHIi9IeoqUpY6GQXP6bH4ZrCTw4e9xfcfjk3Uz2q9+r0kPMHICLAQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "qVEsr7bcara1XYeXsf0n2J0VT2RIdNte7F5A6GDGXWE="
        },
        "tokens_per_usd_hour": 32
      },
      "cheapest": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": null,
        "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
        "ref": "llama3.2:1b",
        "engine": "ollama-host",
        "tokens_per_sec": 20.37,
        "cost_usd_per_million": 8.5986,
        "measured_at": "2026-10-08T15:04:04.375Z",
        "context_tokens": 32768,
        "proof": {
          "node_did": "did:epn:002408011220ab92819c6391e87d553e27cec9a18a52937a84a123bf752d00c2aa2bee0548a8",
          "signature": "eebpktI5uYJizQhh7Sn0Iv+0jvsz9CWBbeHIi9IeoqUpY6GQXP6bH4ZrCTw4e9xfcfjk3Uz2q9+r0kPMHICLAQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "qVEsr7bcara1XYeXsf0n2J0VT2RIdNte7F5A6GDGXWE="
        },
        "tokens_per_usd_hour": 32
      },
      "builds": [
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": null,
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "tokens_per_sec": 20.37,
          "cost_usd_per_million": 8.5986,
          "measured_at": "2026-10-08T15:04:04.375Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:002408011220ab92819c6391e87d553e27cec9a18a52937a84a123bf752d00c2aa2bee0548a8",
            "signature": "eebpktI5uYJizQhh7Sn0Iv+0jvsz9CWBbeHIi9IeoqUpY6GQXP6bH4ZrCTw4e9xfcfjk3Uz2q9+r0kPMHICLAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "qVEsr7bcara1XYeXsf0n2J0VT2RIdNte7F5A6GDGXWE="
          },
          "tokens_per_usd_hour": 32
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "Q4_K_M (4-bit) by lmstudio-community",
          "build_page": "/inferences/phi-4/mini-reasoning/lmstudio-community-q4_k_m-ollama",
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 7.26,
          "cost_usd_per_million": 24.1396,
          "measured_at": "2026-10-08T15:04:04.375Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ab92819c6391e87d553e27cec9a18a52937a84a123bf752d00c2aa2bee0548a8",
            "signature": "gx1LEU02dfkRtRmuI0d2Aawbv054q5xHhItllU2vE6Qh3XqovnbFFbhKsvOdgF9yRgrezd3T4sAcZi7I+BKfCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "O3xHJzTRlWoGTgNmPfzlwPAfucXKmYvQg7+7apV4iiU="
          },
          "tokens_per_usd_hour": 12
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "Q3_K_L (3-bit) by lmstudio-community",
          "build_page": "/inferences/phi-4/mini-reasoning/lmstudio-community-q3_k_l-ollama",
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "tokens_per_sec": 3.42,
          "cost_usd_per_million": 51.1518,
          "measured_at": "2026-10-08T15:04:04.375Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220ab92819c6391e87d553e27cec9a18a52937a84a123bf752d00c2aa2bee0548a8",
            "signature": "jkZhZ1DlFxGRDHhUvtlbWAUYVA4gk5qsmuAVs2J2zHAs0M6+ccw/yJsEagHYtdajmpg180ccgTs0jG1MHJhaBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "93PcLOh/htMZt5b+qnibG7MDO0JlUYRgvS6tN00/h6E="
          },
          "tokens_per_usd_hour": 5
        }
      ],
      "rank": 14,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": null,
        "cpu_all_cores": null,
        "cpu_one_core": null,
        "disk_read": null,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "0c5265",
      "did": "did:epn:0024080112207af079ace316148c9f3e619516325fab90344cf43a0c960cca42eb304a0c5265",
      "name": "Gram ·0c5265",
      "claimed": false,
      "online": false,
      "last_seen": null,
      "version": null,
      "region": null,
      "hardware": {
        "os": "darwin",
        "arch": "arm64",
        "class": null,
        "cpu_model": null,
        "cpu_clock_ghz": null,
        "cpus": 8,
        "ram_gb": 8,
        "memory_bandwidth_gb_s": null,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": null,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": null
      },
      "sheet": null,
      "signed_runs": 8,
      "signed_runs_30d": 0,
      "builds_measured": 8,
      "best_results": 0,
      "first_signed_at": "2026-10-08T07:50:36.302Z",
      "last_signed_at": "2026-10-08T07:50:36.302Z",
      "awake_pct": null,
      "queue": null,
      "download_mib_per_sec": null,
      "uptime_days": [],
      "uptime_pct": null,
      "best_value": null,
      "opened": [
        {
          "ref": "llama3.2:1b",
          "family": "llama-3-2",
          "member": "1b",
          "at": "2026-10-08T07:50:36.302Z"
        },
        {
          "ref": "gemma4:e2b",
          "family": "gemma-4",
          "member": "e2b",
          "at": "2026-10-08T07:50:36.302Z"
        },
        {
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q3_K_M",
          "family": "phi-4",
          "member": "mini-reasoning",
          "at": "2026-10-08T07:50:36.302Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3-VL-4B-Instruct-GGUF:Q4_0",
          "family": "qwen-3",
          "member": "4b",
          "at": "2026-10-08T07:50:36.302Z"
        },
        {
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:Q4_1",
          "family": "qwen-3",
          "member": "4b",
          "at": "2026-10-08T07:50:36.302Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-08",
          "runs": 5
        }
      ],
      "runs_by_family": [
        {
          "family": "phi-4",
          "runs": 3
        },
        {
          "family": "qwen-3",
          "runs": 2
        },
        {
          "family": "llama-3-2",
          "runs": 1
        },
        {
          "family": "musicgen",
          "runs": 1
        },
        {
          "family": "gemma-4",
          "runs": 1
        }
      ],
      "rooms": [
        "cz4QonwjB3lKR0mfHabp",
        "IN_500001"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "opener",
          "label": "First to measure 5 builds"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 15,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": null,
        "cpu_all_cores": null,
        "cpu_one_core": null,
        "disk_read": null,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "9a1db0",
      "did": "did:epn:002408011220b3960d56a82de98ec563fb48bbfc3ca6b4dc788b19545f4315f8f363aa9a1db0",
      "name": "Gram ·9a1db0",
      "claimed": false,
      "online": false,
      "last_seen": null,
      "version": null,
      "region": null,
      "hardware": {
        "os": "windows",
        "arch": "amd64",
        "class": null,
        "cpu_model": null,
        "cpu_clock_ghz": null,
        "cpus": 12,
        "ram_gb": 7.3,
        "memory_bandwidth_gb_s": null,
        "gpu": null,
        "vram_gb": 4,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": null,
        "gpu_count": 1,
        "gpu_org": null,
        "cpu_org": null
      },
      "sheet": null,
      "signed_runs": 6,
      "signed_runs_30d": 0,
      "builds_measured": 6,
      "best_results": 2,
      "first_signed_at": "2026-09-22T10:54:02.460Z",
      "last_signed_at": "2026-09-22T10:54:02.460Z",
      "awake_pct": null,
      "queue": null,
      "download_mib_per_sec": null,
      "uptime_days": [],
      "uptime_pct": null,
      "best_value": {
        "model": "Gemma 4 E2B",
        "page": "/inferences/gemma-4/e2b",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
        "engine": "ollama",
        "tokens_per_sec": 15.33,
        "cost_usd_per_million": 14.2764,
        "measured_at": "2026-09-22T10:54:02.459Z",
        "context_tokens": 131072,
        "proof": {
          "node_did": "did:epn:002408011220b3960d56a82de98ec563fb48bbfc3ca6b4dc788b19545f4315f8f363aa9a1db0",
          "signature": "BKh6EJDNuj5VCyiwsFSToaK8T+f+QRBFAB4GsKdgvONp8CJL68IfgeJ7ADUT0cC8vnBgYAA66/mtlTjz5D3PDg==",
          "signing_payload_b64": "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",
          "payload_sha256": "8mu9Eb6jJxaeXAcubqD5SOFwPqaAdreeRWaSMkJ1Y80="
        },
        "tokens_per_usd_hour": 19
      },
      "opened": [
        {
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "family": "gemma-4",
          "member": "e2b",
          "at": "2026-09-22T10:54:02.460Z"
        },
        {
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
          "family": "qwen-2-5",
          "member": "0.5b",
          "at": "2026-09-22T10:54:02.460Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-09-22",
          "runs": 4
        }
      ],
      "runs_by_family": [
        {
          "family": "other",
          "runs": 3
        },
        {
          "family": "gemma-4",
          "runs": 1
        },
        {
          "family": "musicgen",
          "runs": 1
        },
        {
          "family": "qwen-2-5",
          "runs": 1
        }
      ],
      "rooms": [
        "IN_600001",
        "IN_700108"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "opener",
          "label": "First to measure 2 builds"
        }
      ],
      "fastest": {
        "model": "Gemma 4 E2B",
        "page": "/inferences/gemma-4/e2b",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
        "engine": "ollama",
        "tokens_per_sec": 15.33,
        "cost_usd_per_million": 14.2764,
        "measured_at": "2026-09-22T10:54:02.459Z",
        "context_tokens": 131072,
        "proof": {
          "node_did": "did:epn:002408011220b3960d56a82de98ec563fb48bbfc3ca6b4dc788b19545f4315f8f363aa9a1db0",
          "signature": "BKh6EJDNuj5VCyiwsFSToaK8T+f+QRBFAB4GsKdgvONp8CJL68IfgeJ7ADUT0cC8vnBgYAA66/mtlTjz5D3PDg==",
          "signing_payload_b64": "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",
          "payload_sha256": "8mu9Eb6jJxaeXAcubqD5SOFwPqaAdreeRWaSMkJ1Y80="
        },
        "tokens_per_usd_hour": 19
      },
      "cheapest": {
        "model": "Gemma 4 E2B",
        "page": "/inferences/gemma-4/e2b",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
        "engine": "ollama",
        "tokens_per_sec": 15.33,
        "cost_usd_per_million": 14.2764,
        "measured_at": "2026-09-22T10:54:02.459Z",
        "context_tokens": 131072,
        "proof": {
          "node_did": "did:epn:002408011220b3960d56a82de98ec563fb48bbfc3ca6b4dc788b19545f4315f8f363aa9a1db0",
          "signature": "BKh6EJDNuj5VCyiwsFSToaK8T+f+QRBFAB4GsKdgvONp8CJL68IfgeJ7ADUT0cC8vnBgYAA66/mtlTjz5D3PDg==",
          "signing_payload_b64": "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",
          "payload_sha256": "8mu9Eb6jJxaeXAcubqD5SOFwPqaAdreeRWaSMkJ1Y80="
        },
        "tokens_per_usd_hour": 19
      },
      "builds": [
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama",
          "tokens_per_sec": 15.33,
          "cost_usd_per_million": 14.2764,
          "measured_at": "2026-09-22T10:54:02.459Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220b3960d56a82de98ec563fb48bbfc3ca6b4dc788b19545f4315f8f363aa9a1db0",
            "signature": "BKh6EJDNuj5VCyiwsFSToaK8T+f+QRBFAB4GsKdgvONp8CJL68IfgeJ7ADUT0cC8vnBgYAA66/mtlTjz5D3PDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "8mu9Eb6jJxaeXAcubqD5SOFwPqaAdreeRWaSMkJ1Y80="
          },
          "tokens_per_usd_hour": 19
        },
        {
          "model": "Qwen2.5 0.5B Instruct",
          "page": "/inferences/qwen-2-5/0.5b",
          "build": "Q4_K_M (4-bit) by Qwen",
          "build_page": "/inferences/qwen-2-5/0.5b/qwen-q4_k_m-ollama-148138f0",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 5.75,
          "cost_usd_per_million": 37.5623,
          "measured_at": "2026-09-22T10:54:02.459Z",
          "context_tokens": 32768,
          "proof": {
            "node_did": "did:epn:002408011220b3960d56a82de98ec563fb48bbfc3ca6b4dc788b19545f4315f8f363aa9a1db0",
            "signature": "bTSzZMkpKaUGtatlxpE3nu4sGTK+PHYPBYUiLpW5XtP3Bytz5P2LvvvYvWYSprOWRAcjUL31UuqSTwSjbQaaBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "/KiVhGDZJmdErAW8JWT09zN/zcJrgeJ181/HdZQJENg="
          },
          "tokens_per_usd_hour": 7
        }
      ],
      "rank": 16,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": null,
        "cpu_all_cores": null,
        "cpu_one_core": null,
        "disk_read": null,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "92d334",
      "did": "did:epn:002408011220b36c12ce53e9096b1e6574d05f18f09ba07385e81dc933d849a1cc239392d334",
      "name": "Gram ·92d334",
      "claimed": true,
      "online": false,
      "last_seen": "2026-11-15T23:40:28.628Z",
      "version": "46aa233",
      "region": {
        "code": "IN_752115",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "ios",
        "arch": "arm64",
        "class": "ios/arm64/none/4g",
        "cpu_model": null,
        "cpu_clock_ghz": null,
        "cpus": 5,
        "ram_gb": null,
        "memory_bandwidth_gb_s": 2.5,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 1460.1,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": null
      },
      "sheet": {
        "cpu": {
          "model": null,
          "model_raw": null,
          "clock_ghz": null,
          "cores_logical": 5,
          "cores_physical": null,
          "threads_per_core": null,
          "sockets": null,
          "l2_kb": null,
          "l3_mb": null,
          "single_core_hashes_per_sec": 7,
          "multi_core_hashes_per_sec": 55.9,
          "sustained_ratio": 3.6,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-14T16:34:58.410Z",
          "measured_in": "IN_752115",
          "signed": true
        },
        "memory": {
          "total_gb": null,
          "bandwidth_gb_s": 2.5,
          "latency_ns": 150.9,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-14T16:34:59.487Z",
          "measured_in": "IN_752115",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 1460.1,
          "write_mb_s": 828.4,
          "random_read_iops": 108925,
          "fs_type": null,
          "device_model": null,
          "measured_at": "2026-11-14T16:35:00.458Z",
          "measured_in": "IN_752115",
          "signed": true
        }
      },
      "signed_runs": 5,
      "signed_runs_30d": 5,
      "builds_measured": 5,
      "best_results": 0,
      "first_signed_at": "2026-11-06T21:35:00.335Z",
      "last_signed_at": "2026-11-14T16:35:24.459Z",
      "awake_pct": 0.2,
      "queue": {
        "measured": 0,
        "failed": 0,
        "queued": 1272
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-11-05",
          "hours": 1
        },
        {
          "date": "2026-11-06",
          "hours": 5
        },
        {
          "date": "2026-11-07",
          "hours": 14
        },
        {
          "date": "2026-11-08",
          "hours": 11
        },
        {
          "date": "2026-11-09",
          "hours": 9
        },
        {
          "date": "2026-11-10",
          "hours": 12
        },
        {
          "date": "2026-11-11",
          "hours": 6
        },
        {
          "date": "2026-11-12",
          "hours": 16
        },
        {
          "date": "2026-11-13",
          "hours": 13
        },
        {
          "date": "2026-11-14",
          "hours": 18
        },
        {
          "date": "2026-11-15",
          "hours": 10
        }
      ],
      "uptime_pct": 43.6,
      "best_value": null,
      "opened": [],
      "runs_by_day": [
        {
          "date": "2026-11-06",
          "runs": 3
        },
        {
          "date": "2026-11-07",
          "runs": 44
        },
        {
          "date": "2026-11-08",
          "runs": 30
        },
        {
          "date": "2026-11-09",
          "runs": 36
        },
        {
          "date": "2026-11-10",
          "runs": 37
        },
        {
          "date": "2026-11-11",
          "runs": 20
        },
        {
          "date": "2026-11-12",
          "runs": 31
        },
        {
          "date": "2026-11-13",
          "runs": 38
        },
        {
          "date": "2026-11-14",
          "runs": 30
        }
      ],
      "runs_by_family": [
        {
          "family": "other",
          "runs": 3
        },
        {
          "family": "qwen-2-5",
          "runs": 1
        },
        {
          "family": "gpt-oss",
          "runs": 1
        }
      ],
      "rooms": [
        "IN_752115"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 17,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 56,
        "cpu_all_cores": 4,
        "cpu_one_core": 0,
        "disk_read": 21,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "53b229",
      "did": "did:epn:0024080112201844bc3835f475fd8afa64d04ddcb9a1e0087bb40a31973d91b193e6c953b229",
      "name": "Gram ·53b229",
      "claimed": false,
      "online": false,
      "last_seen": "2026-11-01T21:45:27.889Z",
      "version": "efaaa4d",
      "region": {
        "code": "IN_560037",
        "city": "Bengaluru",
        "state": "Karnataka"
      },
      "hardware": {
        "os": "windows",
        "arch": "x86_64",
        "class": "windows/amd64/nvidia/16g",
        "cpu_model": "11th Gen Intel Core i7-11600H",
        "cpu_clock_ghz": null,
        "cpus": 12,
        "ram_gb": 15.729339599609375,
        "memory_bandwidth_gb_s": 4.2,
        "gpu": "NVIDIA GeForce RTX 3050 Laptop GPU",
        "vram_gb": 4,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 1409.8,
        "gpu_count": 1,
        "gpu_org": "nvidia",
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "11th Gen Intel Core i7-11600H",
          "model_raw": "11th Gen Intel Core i7-11600H",
          "clock_ghz": null,
          "cores_logical": 12,
          "cores_physical": null,
          "threads_per_core": null,
          "sockets": null,
          "l2_kb": null,
          "l3_mb": null,
          "single_core_hashes_per_sec": 365.3,
          "multi_core_hashes_per_sec": 3482.5,
          "sustained_ratio": 0.9,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-01T21:15:49.919Z",
          "measured_in": "IN_560037",
          "signed": true
        },
        "memory": {
          "total_gb": 15.7,
          "bandwidth_gb_s": 4.2,
          "latency_ns": 44.2,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-01T21:15:50.572Z",
          "measured_in": "IN_560037",
          "signed": true
        },
        "gpu": {
          "name": "NVIDIA GeForce RTX 3050 Laptop GPU",
          "memory_gb": 4,
          "fp16_tflops": 0,
          "watts": null,
          "tflops_per_watt": null,
          "method": null,
          "measured_at": "2026-09-11T19:24:20.219Z",
          "measured_in": null,
          "signed": true
        },
        "storage": {
          "read_mb_s": 1409.8,
          "write_mb_s": 340.5,
          "random_read_iops": 46320,
          "fs_type": null,
          "device_model": null,
          "measured_at": "2026-11-01T21:15:51.083Z",
          "measured_in": "IN_560037",
          "signed": true
        }
      },
      "signed_runs": 5,
      "signed_runs_30d": 5,
      "builds_measured": 5,
      "best_results": 3,
      "first_signed_at": "2026-10-28T01:55:27.489Z",
      "last_signed_at": "2026-11-01T21:35:20.666Z",
      "awake_pct": 68.7,
      "queue": {
        "measured": 0,
        "failed": 5,
        "queued": 159
      },
      "download_mib_per_sec": 6.6,
      "uptime_days": [
        {
          "date": "2026-10-30",
          "hours": 1
        },
        {
          "date": "2026-11-01",
          "hours": 1
        }
      ],
      "uptime_pct": 4.2,
      "best_value": {
        "model": "Phi 4 Mini-Reasoning",
        "page": "/inferences/phi-4/mini-reasoning",
        "build": "IQ3_XXS (3-bit) by unsloth",
        "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq3_xxs-ollama",
        "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
        "engine": "ollama-host",
        "tokens_per_sec": 41.79,
        "cost_usd_per_million": 5.1926,
        "measured_at": "2026-11-01T21:35:20.666Z",
        "context_tokens": 4096,
        "proof": {
          "node_did": "did:epn:0024080112201844bc3835f475fd8afa64d04ddcb9a1e0087bb40a31973d91b193e6c953b229",
          "signature": "gDRvtw7fPU4gl1dM/zjUVcGGX2ETeMmCXTafGYhS0gvdC+3fqBeilyANhBZ7ZHWl8xbPqPNSAW1iXDl/JmamBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "WpmA3Ye/biYhvXQ85EuZ+5tdogk9mndQsW5Sm2HecZ4="
        },
        "tokens_per_usd_hour": 53
      },
      "opened": [
        {
          "ref": "hf.co/mradermacher/Airavata-GGUF:Q4_K_M",
          "family": "airavata",
          "member": "7b",
          "at": "2026-11-01T21:35:20.666Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-28",
          "runs": 4
        },
        {
          "date": "2026-10-30",
          "runs": 8
        },
        {
          "date": "2026-11-01",
          "runs": 10
        }
      ],
      "runs_by_family": [
        {
          "family": "phi-4",
          "runs": 3
        },
        {
          "family": "gemma-4",
          "runs": 1
        },
        {
          "family": "airavata",
          "runs": 1
        }
      ],
      "rooms": [
        "IN_560001",
        "IN_560037"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "opener",
          "label": "First to measure 1 build"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "Phi 4 Mini-Reasoning",
        "page": "/inferences/phi-4/mini-reasoning",
        "build": "IQ3_XXS (3-bit) by unsloth",
        "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq3_xxs-ollama",
        "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
        "engine": "ollama-host",
        "tokens_per_sec": 41.79,
        "cost_usd_per_million": 5.1926,
        "measured_at": "2026-11-01T21:35:20.666Z",
        "context_tokens": 4096,
        "proof": {
          "node_did": "did:epn:0024080112201844bc3835f475fd8afa64d04ddcb9a1e0087bb40a31973d91b193e6c953b229",
          "signature": "gDRvtw7fPU4gl1dM/zjUVcGGX2ETeMmCXTafGYhS0gvdC+3fqBeilyANhBZ7ZHWl8xbPqPNSAW1iXDl/JmamBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "WpmA3Ye/biYhvXQ85EuZ+5tdogk9mndQsW5Sm2HecZ4="
        },
        "tokens_per_usd_hour": 53
      },
      "cheapest": {
        "model": "Phi 4 Mini-Reasoning",
        "page": "/inferences/phi-4/mini-reasoning",
        "build": "IQ3_XXS (3-bit) by unsloth",
        "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq3_xxs-ollama",
        "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
        "engine": "ollama-host",
        "tokens_per_sec": 41.79,
        "cost_usd_per_million": 5.1926,
        "measured_at": "2026-11-01T21:35:20.666Z",
        "context_tokens": 4096,
        "proof": {
          "node_did": "did:epn:0024080112201844bc3835f475fd8afa64d04ddcb9a1e0087bb40a31973d91b193e6c953b229",
          "signature": "gDRvtw7fPU4gl1dM/zjUVcGGX2ETeMmCXTafGYhS0gvdC+3fqBeilyANhBZ7ZHWl8xbPqPNSAW1iXDl/JmamBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "WpmA3Ye/biYhvXQ85EuZ+5tdogk9mndQsW5Sm2HecZ4="
        },
        "tokens_per_usd_hour": 53
      },
      "builds": [
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "IQ3_XXS (3-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq3_xxs-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "tokens_per_sec": 41.79,
          "cost_usd_per_million": 5.1926,
          "measured_at": "2026-11-01T21:35:20.666Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:0024080112201844bc3835f475fd8afa64d04ddcb9a1e0087bb40a31973d91b193e6c953b229",
            "signature": "gDRvtw7fPU4gl1dM/zjUVcGGX2ETeMmCXTafGYhS0gvdC+3fqBeilyANhBZ7ZHWl8xbPqPNSAW1iXDl/JmamBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "WpmA3Ye/biYhvXQ85EuZ+5tdogk9mndQsW5Sm2HecZ4="
          },
          "tokens_per_usd_hour": 53
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "tokens_per_sec": 28.05,
          "cost_usd_per_million": 7.8028,
          "measured_at": "2026-11-01T21:35:20.666Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:0024080112201844bc3835f475fd8afa64d04ddcb9a1e0087bb40a31973d91b193e6c953b229",
            "signature": "Gpwv/X8hmzFtV/ZcQ2e7QmNchYVdBaKyv/X31LdiK/QVFcF7KUVK6fxUF5u6ZrXQbQjpb91fZtsGsTlBmIoFAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "6A5NLcqkeGMVjP7JR875olLeAK09QSJT1ezPwpRxHi4="
          },
          "tokens_per_usd_hour": 36
        },
        {
          "model": "Phi 4 Mini-Reasoning",
          "page": "/inferences/phi-4/mini-reasoning",
          "build": "IQ1_S (1-bit) by unsloth",
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq1_s-ollama",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ1_S",
          "engine": "ollama-host",
          "tokens_per_sec": 16.34,
          "cost_usd_per_million": 13.2553,
          "measured_at": "2026-11-01T21:35:20.666Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:0024080112201844bc3835f475fd8afa64d04ddcb9a1e0087bb40a31973d91b193e6c953b229",
            "signature": "zCysT6iezXgDmBvHiRfSiVeRLj0yHgn6apQPaWwFrUaEurTZUetY19HtFKp0ZjGPunf2THurbziRV19oS8H8BQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "t8mOB8pC2oyq6NIBQ7aeeHLDY2QGQregUa2R22liNGY="
          },
          "tokens_per_usd_hour": 21
        }
      ],
      "rank": 18,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 67,
        "cpu_all_cores": 75,
        "cpu_one_core": 68,
        "disk_read": 18,
        "download": 20
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "970094",
      "did": "did:epn:002408011220b693cfea518e20e0c1d424fc7843fa4f8f1ac04691f5a6fbd3d7c03bfc970094",
      "name": "Gram ·970094",
      "claimed": true,
      "online": false,
      "last_seen": "2026-11-15T11:47:30.636Z",
      "version": "f3f806e",
      "region": {
        "code": "IN_752115",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "ios",
        "arch": "arm64",
        "class": "ios/arm64/none/4g",
        "cpu_model": null,
        "cpu_clock_ghz": null,
        "cpus": 6,
        "ram_gb": null,
        "memory_bandwidth_gb_s": 1.1,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 2225.7,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": null
      },
      "sheet": {
        "cpu": {
          "model": null,
          "model_raw": null,
          "clock_ghz": null,
          "cores_logical": 6,
          "cores_physical": null,
          "threads_per_core": null,
          "sockets": null,
          "l2_kb": null,
          "l3_mb": null,
          "single_core_hashes_per_sec": 50.1,
          "multi_core_hashes_per_sec": 99.9,
          "sustained_ratio": 0.8,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-11T05:56:17.871Z",
          "measured_in": "IN_752115",
          "signed": true
        },
        "memory": {
          "total_gb": null,
          "bandwidth_gb_s": 1.1,
          "latency_ns": 81.3,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-11T05:56:18.779Z",
          "measured_in": "IN_752115",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 2225.7,
          "write_mb_s": 98.9,
          "random_read_iops": 17586,
          "fs_type": null,
          "device_model": null,
          "measured_at": "2026-11-11T05:56:19.980Z",
          "measured_in": "IN_752115",
          "signed": true
        }
      },
      "signed_runs": 4,
      "signed_runs_30d": 4,
      "builds_measured": 4,
      "best_results": 0,
      "first_signed_at": "2026-11-08T03:55:52.519Z",
      "last_signed_at": "2026-11-11T07:52:49.072Z",
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 0,
        "queued": 249
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-11-06",
          "hours": 1
        },
        {
          "date": "2026-11-07",
          "hours": 4
        },
        {
          "date": "2026-11-08",
          "hours": 5
        },
        {
          "date": "2026-11-09",
          "hours": 6
        },
        {
          "date": "2026-11-10",
          "hours": 5
        },
        {
          "date": "2026-11-11",
          "hours": 4
        },
        {
          "date": "2026-11-12",
          "hours": 5
        },
        {
          "date": "2026-11-13",
          "hours": 6
        },
        {
          "date": "2026-11-14",
          "hours": 2
        },
        {
          "date": "2026-11-15",
          "hours": 4
        }
      ],
      "uptime_pct": 17.5,
      "best_value": null,
      "opened": [
        {
          "ref": "claude-opus-5",
          "family": null,
          "member": null,
          "at": "2026-11-11T07:52:49.072Z"
        },
        {
          "ref": "claude-sonnet-5",
          "family": null,
          "member": null,
          "at": "2026-11-11T07:52:49.072Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-11-08",
          "runs": 6
        },
        {
          "date": "2026-11-09",
          "runs": 6
        },
        {
          "date": "2026-11-10",
          "runs": 5
        },
        {
          "date": "2026-11-11",
          "runs": 4
        }
      ],
      "runs_by_family": [
        {
          "family": "other",
          "runs": 3
        },
        {
          "family": "qwen-2-5",
          "runs": 1
        }
      ],
      "rooms": [
        "IN_752115"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "opener",
          "label": "First to measure 2 builds"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 19,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 19,
        "cpu_all_cores": 7,
        "cpu_one_core": 7,
        "disk_read": 39,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "c76d30",
      "did": "did:epn:002408011220d79e72fae9f23f0437b58a3411e6bdb14e5c77f568cc3bdf6622534463c76d30",
      "name": "Gram ·c76d30",
      "claimed": false,
      "online": false,
      "last_seen": "2026-11-03T00:30:53.375Z",
      "version": "bdc85f4",
      "region": {
        "code": "TW_110",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/nvidia/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2,
        "cpus": 4,
        "ram_gb": 31.346599578857422,
        "memory_bandwidth_gb_s": 2.4,
        "gpu": "Tesla T4",
        "vram_gb": 14.6,
        "gpu_tflops_fp16": 21.6,
        "disk_read_mb_s": 4293.1,
        "gpu_count": 2,
        "gpu_org": "nvidia",
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
          "clock_ghz": 2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 1024,
          "l3_mb": 38.5,
          "single_core_hashes_per_sec": 312.8,
          "multi_core_hashes_per_sec": 618.9,
          "sustained_ratio": 1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-02T21:33:01.078Z",
          "measured_in": "TW_110",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 2.4,
          "latency_ns": 21.6,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-02T21:33:01.673Z",
          "measured_in": "TW_110",
          "signed": true
        },
        "gpu": {
          "name": "Tesla T4",
          "memory_gb": 14.6,
          "fp16_tflops": 21.6,
          "watts": 59.1,
          "tflops_per_watt": 0.365,
          "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
          "measured_at": "2026-11-02T21:43:16.805Z",
          "measured_in": "TW_110",
          "signed": true
        },
        "storage": {
          "read_mb_s": 4293.1,
          "write_mb_s": 60.4,
          "random_read_iops": 514568,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-11-02T21:33:02.973Z",
          "measured_in": "TW_110",
          "signed": true
        }
      },
      "signed_runs": 4,
      "signed_runs_30d": 4,
      "builds_measured": 4,
      "best_results": 3,
      "first_signed_at": "2026-11-02T21:51:51.816Z",
      "last_signed_at": "2026-11-03T00:21:51.818Z",
      "awake_pct": 100,
      "queue": {
        "measured": 3,
        "failed": 0,
        "queued": 1753
      },
      "download_mib_per_sec": 83.9,
      "uptime_days": [
        {
          "date": "2026-11-02",
          "hours": 3
        },
        {
          "date": "2026-11-03",
          "hours": 1
        }
      ],
      "uptime_pct": 8.3,
      "best_value": {
        "model": "Granite 4.0 Speech-1B",
        "page": "/inferences/granite-4-0/speech-1b",
        "build": "Q4_K_M (4-bit) by ibm-granite",
        "build_page": "/inferences/granite-4-0/speech-1b/ibm-granite-q4_k_m-ollama",
        "ref": "hf.co/ibm-granite/granite-4.0-1b-speech-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 129.7,
        "cost_usd_per_million": 1.0444,
        "measured_at": "2026-11-03T00:21:51.818Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220d79e72fae9f23f0437b58a3411e6bdb14e5c77f568cc3bdf6622534463c76d30",
          "signature": "Fm1ATdm6uyWZLgWqRl0N+H8lLhBqiphWUjfNmJc4WMOOjLCX5u/Npv3riQbav1c527iIF0XBBy9zxuIBWvcbDA==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTI5LjcwNDAyMDA4OTYzNjY5LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5MzY2NTMxMTgxODMzODEwMSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMGQ3OWU3MmZhZTlmMjNmMDQzN2I1OGEzNDExZTZiZGIxNGU1Yzc3ZjU2OGNjM2JkZjY2MjI1MzQ0NjNjNzZkMzAiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L252aWRpYS8zMmciLCJtb2RlbCI6ImhmLmNvL2libS1ncmFuaXRlL2dyYW5pdGUtNC4wLTFiLXNwZWVjaC1HR1VGOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo1MSwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjo5Ljg5MTc1MTk5OSwidG90YWxfdG9rZW5zIjoxMjgzfX0=",
          "payload_sha256": "67asdisH59BWTjznW2/Hb+qCNI3+1+RDD8WWtarvqsI="
        },
        "tokens_per_usd_hour": 266
      },
      "opened": [
        {
          "ref": "hf.co/ibm-granite/granite-4.0-1b-speech-GGUF:Q4_K_M",
          "family": "granite-4-0",
          "member": "speech-1b",
          "at": "2026-11-03T00:21:51.818Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-11-02",
          "runs": 28
        },
        {
          "date": "2026-11-03",
          "runs": 9
        }
      ],
      "runs_by_family": [
        {
          "family": "gemma-4",
          "runs": 1
        },
        {
          "family": "gpt-oss",
          "runs": 1
        },
        {
          "family": "qwen-3-6",
          "runs": 1
        },
        {
          "family": "granite-4-0",
          "runs": 1
        }
      ],
      "rooms": [
        "TW_110"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "opener",
          "label": "First to measure 1 build"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "Granite 4.0 Speech-1B",
        "page": "/inferences/granite-4-0/speech-1b",
        "build": "Q4_K_M (4-bit) by ibm-granite",
        "build_page": "/inferences/granite-4-0/speech-1b/ibm-granite-q4_k_m-ollama",
        "ref": "hf.co/ibm-granite/granite-4.0-1b-speech-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 129.7,
        "cost_usd_per_million": 1.0444,
        "measured_at": "2026-11-03T00:21:51.818Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220d79e72fae9f23f0437b58a3411e6bdb14e5c77f568cc3bdf6622534463c76d30",
          "signature": "Fm1ATdm6uyWZLgWqRl0N+H8lLhBqiphWUjfNmJc4WMOOjLCX5u/Npv3riQbav1c527iIF0XBBy9zxuIBWvcbDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "67asdisH59BWTjznW2/Hb+qCNI3+1+RDD8WWtarvqsI="
        },
        "tokens_per_usd_hour": 266
      },
      "cheapest": {
        "model": "Granite 4.0 Speech-1B",
        "page": "/inferences/granite-4-0/speech-1b",
        "build": "Q4_K_M (4-bit) by ibm-granite",
        "build_page": "/inferences/granite-4-0/speech-1b/ibm-granite-q4_k_m-ollama",
        "ref": "hf.co/ibm-granite/granite-4.0-1b-speech-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "tokens_per_sec": 129.7,
        "cost_usd_per_million": 1.0444,
        "measured_at": "2026-11-03T00:21:51.818Z",
        "context_tokens": null,
        "proof": {
          "node_did": "did:epn:002408011220d79e72fae9f23f0437b58a3411e6bdb14e5c77f568cc3bdf6622534463c76d30",
          "signature": "Fm1ATdm6uyWZLgWqRl0N+H8lLhBqiphWUjfNmJc4WMOOjLCX5u/Npv3riQbav1c527iIF0XBBy9zxuIBWvcbDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "67asdisH59BWTjznW2/Hb+qCNI3+1+RDD8WWtarvqsI="
        },
        "tokens_per_usd_hour": 266
      },
      "builds": [
        {
          "model": "Granite 4.0 Speech-1B",
          "page": "/inferences/granite-4-0/speech-1b",
          "build": "Q4_K_M (4-bit) by ibm-granite",
          "build_page": "/inferences/granite-4-0/speech-1b/ibm-granite-q4_k_m-ollama",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-speech-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 129.7,
          "cost_usd_per_million": 1.0444,
          "measured_at": "2026-11-03T00:21:51.818Z",
          "context_tokens": null,
          "proof": {
            "node_did": "did:epn:002408011220d79e72fae9f23f0437b58a3411e6bdb14e5c77f568cc3bdf6622534463c76d30",
            "signature": "Fm1ATdm6uyWZLgWqRl0N+H8lLhBqiphWUjfNmJc4WMOOjLCX5u/Npv3riQbav1c527iIF0XBBy9zxuIBWvcbDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "67asdisH59BWTjznW2/Hb+qCNI3+1+RDD8WWtarvqsI="
          },
          "tokens_per_usd_hour": 266
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "tokens_per_sec": 74.18,
          "cost_usd_per_million": 1.8572,
          "measured_at": "2026-11-03T00:21:51.818Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220d79e72fae9f23f0437b58a3411e6bdb14e5c77f568cc3bdf6622534463c76d30",
            "signature": "+zNCrJNhpfdBojlMeA+Mp0j/4evVbY618zdYeDdYAGc85a9yuXO3x5EXwQGwZPUP5oF04a5qgpuMOxqPHp4wAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "dVGqqLD+hVl/M/Km3YocSTqlIWfZVsI88YNfyrlk8ic="
          },
          "tokens_per_usd_hour": 150
        },
        {
          "model": "gpt-oss 20B",
          "page": "/inferences/gpt-oss/20b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gpt-oss/20b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "tokens_per_sec": 52.39,
          "cost_usd_per_million": 2.7409,
          "measured_at": "2026-11-03T00:21:51.818Z",
          "context_tokens": 65536,
          "proof": {
            "node_did": "did:epn:002408011220d79e72fae9f23f0437b58a3411e6bdb14e5c77f568cc3bdf6622534463c76d30",
            "signature": "ZlROyZ5tTzYRW8z97CG94RXvxMBBG6cWEdV5YWlAoXc9CjyiqOhhhaTpTC0n446Toubz3OXNrwbYfS9arj9GAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "jEwoDxq9KJMQ9zYLF0Zr40bMTfI90tVTbe4Z8vZUN9E="
          },
          "tokens_per_usd_hour": 101
        }
      ],
      "rank": 20,
      "of": 36,
      "percentiles": {
        "gpu_tflops": 20,
        "memory_bandwidth": 52,
        "cpu_all_cores": 61,
        "cpu_one_core": 64,
        "disk_read": 82,
        "download": 70
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "e189d2",
      "did": "did:epn:002408011220cf803b3f11052307765669b9d23c5c214b142184538b81d7b2779d8cd1e189d2",
      "name": "Gram ·e189d2",
      "claimed": false,
      "online": false,
      "last_seen": "2026-11-03T02:47:59.260Z",
      "version": "62230a4",
      "region": {
        "code": "NL_9711",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/nvidia/256g",
        "cpu_model": "AMD EPYC 9B45",
        "cpu_clock_ghz": null,
        "cpus": 48,
        "ram_gb": 176.88478088378906,
        "memory_bandwidth_gb_s": 26.2,
        "gpu": "NVIDIA RTX PRO 6000 Blackwell Server Edition",
        "vram_gb": 95,
        "gpu_tflops_fp16": 372.6,
        "disk_read_mb_s": 15977,
        "gpu_count": 1,
        "gpu_org": "nvidia",
        "cpu_org": "amd"
      },
      "sheet": {
        "cpu": {
          "model": "AMD EPYC 9B45",
          "model_raw": "AMD EPYC 9B45",
          "clock_ghz": null,
          "cores_logical": 48,
          "cores_physical": 24,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 1024,
          "l3_mb": 32,
          "single_core_hashes_per_sec": 1937.9,
          "multi_core_hashes_per_sec": 70559.4,
          "sustained_ratio": 1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-03T02:23:41.449Z",
          "measured_in": "NL_9711",
          "signed": true
        },
        "memory": {
          "total_gb": 176.9,
          "bandwidth_gb_s": 26.2,
          "latency_ns": 12,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-03T02:23:41.985Z",
          "measured_in": "NL_9711",
          "signed": true
        },
        "gpu": {
          "name": "NVIDIA RTX PRO 6000 Blackwell Server Edition",
          "memory_gb": 95,
          "fp16_tflops": 372.6,
          "watts": 39.8,
          "tflops_per_watt": 9.353,
          "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
          "measured_at": "2026-11-03T02:33:49.172Z",
          "measured_in": "NL_9711",
          "signed": true
        },
        "storage": {
          "read_mb_s": 15977,
          "write_mb_s": 36.1,
          "random_read_iops": 1825761,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-11-03T02:23:43.974Z",
          "measured_in": "NL_9711",
          "signed": true
        }
      },
      "signed_runs": 3,
      "signed_runs_30d": 2,
      "builds_measured": 3,
      "best_results": 1,
      "first_signed_at": "2026-11-03T02:42:33.308Z",
      "last_signed_at": "2026-11-03T02:42:33.308Z",
      "awake_pct": 100,
      "queue": {
        "measured": 2,
        "failed": 0,
        "queued": 1054
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-11-03",
          "hours": 1
        }
      ],
      "uptime_pct": 4.2,
      "best_value": {
        "model": "Gemma 4 E2B",
        "page": "/inferences/gemma-4/e2b",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
        "engine": "ollama-host",
        "tokens_per_sec": 291.17,
        "cost_usd_per_million": 2.0053,
        "measured_at": "2026-11-03T02:42:33.308Z",
        "context_tokens": 131072,
        "proof": {
          "node_did": "did:epn:002408011220cf803b3f11052307765669b9d23c5c214b142184538b81d7b2779d8cd1e189d2",
          "signature": "v0KKPp0Se3QbAB8vgwMmja9knVDNuEvglNHE4SuN2zcR3uZsX7LvcPmKUBW5Gev5eCacSafd/gCCa3jQGtf/BA==",
          "signing_payload_b64": "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",
          "payload_sha256": "KGFGnLsCiMhio27cBxMKafXeCyrjOYV+mUk/A0nZ+WM="
        },
        "tokens_per_usd_hour": 139
      },
      "opened": [],
      "runs_by_day": [
        {
          "date": "2026-11-03",
          "runs": 3
        }
      ],
      "runs_by_family": [
        {
          "family": "glm-4-7",
          "runs": 1
        },
        {
          "family": "gemma-4",
          "runs": 1
        },
        {
          "family": "embeddinggemma",
          "runs": 1
        }
      ],
      "rooms": [],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "Gemma 4 E2B",
        "page": "/inferences/gemma-4/e2b",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
        "engine": "ollama-host",
        "tokens_per_sec": 291.17,
        "cost_usd_per_million": 2.0053,
        "measured_at": "2026-11-03T02:42:33.308Z",
        "context_tokens": 131072,
        "proof": {
          "node_did": "did:epn:002408011220cf803b3f11052307765669b9d23c5c214b142184538b81d7b2779d8cd1e189d2",
          "signature": "v0KKPp0Se3QbAB8vgwMmja9knVDNuEvglNHE4SuN2zcR3uZsX7LvcPmKUBW5Gev5eCacSafd/gCCa3jQGtf/BA==",
          "signing_payload_b64": "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",
          "payload_sha256": "KGFGnLsCiMhio27cBxMKafXeCyrjOYV+mUk/A0nZ+WM="
        },
        "tokens_per_usd_hour": 139
      },
      "cheapest": {
        "model": "Gemma 4 E2B",
        "page": "/inferences/gemma-4/e2b",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
        "engine": "ollama-host",
        "tokens_per_sec": 291.17,
        "cost_usd_per_million": 2.0053,
        "measured_at": "2026-11-03T02:42:33.308Z",
        "context_tokens": 131072,
        "proof": {
          "node_did": "did:epn:002408011220cf803b3f11052307765669b9d23c5c214b142184538b81d7b2779d8cd1e189d2",
          "signature": "v0KKPp0Se3QbAB8vgwMmja9knVDNuEvglNHE4SuN2zcR3uZsX7LvcPmKUBW5Gev5eCacSafd/gCCa3jQGtf/BA==",
          "signing_payload_b64": "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",
          "payload_sha256": "KGFGnLsCiMhio27cBxMKafXeCyrjOYV+mUk/A0nZ+WM="
        },
        "tokens_per_usd_hour": 139
      },
      "builds": [
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "tokens_per_sec": 291.17,
          "cost_usd_per_million": 2.0053,
          "measured_at": "2026-11-03T02:42:33.308Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220cf803b3f11052307765669b9d23c5c214b142184538b81d7b2779d8cd1e189d2",
            "signature": "v0KKPp0Se3QbAB8vgwMmja9knVDNuEvglNHE4SuN2zcR3uZsX7LvcPmKUBW5Gev5eCacSafd/gCCa3jQGtf/BA==",
            "signing_payload_b64": "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",
            "payload_sha256": "KGFGnLsCiMhio27cBxMKafXeCyrjOYV+mUk/A0nZ+WM="
          },
          "tokens_per_usd_hour": 139
        }
      ],
      "rank": 21,
      "of": 36,
      "percentiles": {
        "gpu_tflops": 100,
        "memory_bandwidth": 93,
        "cpu_all_cores": 100,
        "cpu_one_core": 89,
        "disk_read": 100,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "198875",
      "did": "did:epn:002408011220e3153ea1d5d2ac56f8281571a1dbbbca95fb3384035a5f32060d65d5ce198875",
      "name": "Gram ·198875",
      "claimed": true,
      "online": false,
      "last_seen": "2026-10-30T11:31:57.821Z",
      "version": "dc8df19",
      "region": {
        "code": "US_29415",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2.2,
        "cpus": 4,
        "ram_gb": 31.346603393554688,
        "memory_bandwidth_gb_s": 1.2,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 2182.7,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
          "clock_ghz": 2.2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 256,
          "l3_mb": 55,
          "single_core_hashes_per_sec": 181.6,
          "multi_core_hashes_per_sec": 452.5,
          "sustained_ratio": 1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-10-30T06:04:04.572Z",
          "measured_in": "US_29415",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 1.2,
          "latency_ns": 143.6,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-10-30T06:04:05.450Z",
          "measured_in": "US_29415",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 2182.7,
          "write_mb_s": 156.8,
          "random_read_iops": 274587,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-10-30T06:04:06.135Z",
          "measured_in": "US_29415",
          "signed": true
        }
      },
      "signed_runs": 3,
      "signed_runs_30d": 3,
      "builds_measured": 3,
      "best_results": 0,
      "first_signed_at": "2026-10-30T00:27:50.827Z",
      "last_signed_at": "2026-10-30T11:22:50.828Z",
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 0,
        "queued": 1377
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-30",
          "hours": 12
        }
      ],
      "uptime_pct": 50,
      "best_value": null,
      "opened": [],
      "runs_by_day": [
        {
          "date": "2026-10-30",
          "runs": 89
        }
      ],
      "runs_by_family": [
        {
          "family": "llama-3-2",
          "runs": 1
        },
        {
          "family": "glm-4-7",
          "runs": 1
        },
        {
          "family": "other",
          "runs": 1
        }
      ],
      "rooms": [
        "US_29415"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 22,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 22,
        "cpu_all_cores": 43,
        "cpu_one_core": 36,
        "disk_read": 32,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "fe1194",
      "did": "did:epn:002408011220e646f7fab0276df0cdd509b9d442e0c17da93356162faaac0570668334fe1194",
      "name": "Gram ·fe1194",
      "claimed": true,
      "online": false,
      "last_seen": "2026-10-30T11:30:10.696Z",
      "version": "dc8df19",
      "region": {
        "code": "US_97058",
        "city": "The Dalles",
        "state": "Oregon"
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2.2,
        "cpus": 4,
        "ram_gb": 31.346599578857422,
        "memory_bandwidth_gb_s": 0.5,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 1775.2,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
          "clock_ghz": 2.2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 256,
          "l3_mb": 55,
          "single_core_hashes_per_sec": 152.6,
          "multi_core_hashes_per_sec": 422,
          "sustained_ratio": 1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-10-30T06:02:16.551Z",
          "measured_in": "US_97058",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 0.5,
          "latency_ns": 183.1,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-10-30T06:02:17.517Z",
          "measured_in": "US_97058",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 1775.2,
          "write_mb_s": 71.7,
          "random_read_iops": 251738,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-10-30T06:02:18.721Z",
          "measured_in": "US_97058",
          "signed": true
        }
      },
      "signed_runs": 3,
      "signed_runs_30d": 3,
      "builds_measured": 3,
      "best_results": 0,
      "first_signed_at": "2026-10-30T00:31:25.081Z",
      "last_signed_at": "2026-10-30T11:21:25.081Z",
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 0,
        "queued": 1377
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-30",
          "hours": 12
        }
      ],
      "uptime_pct": 50,
      "best_value": null,
      "opened": [],
      "runs_by_day": [
        {
          "date": "2026-10-30",
          "runs": 89
        }
      ],
      "runs_by_family": [
        {
          "family": "llama-3-2",
          "runs": 1
        },
        {
          "family": "glm-4-7",
          "runs": 1
        },
        {
          "family": "other",
          "runs": 1
        }
      ],
      "rooms": [
        "US_97058"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 23,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 11,
        "cpu_all_cores": 32,
        "cpu_one_core": 25,
        "disk_read": 25,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "12954d",
      "did": "did:epn:002408011220df04c29e6a0ad2ef0dc554f22fdbb03346428fde27cc62255d9dd9c61412954d",
      "name": "Gram ·12954d",
      "claimed": false,
      "online": false,
      "last_seen": "2026-10-26T10:47:42.146Z",
      "version": "64a4d32",
      "region": null,
      "hardware": {
        "os": "windows",
        "arch": "x86_64",
        "class": "windows/amd64/nvidia/8g",
        "cpu_model": null,
        "cpu_clock_ghz": null,
        "cpus": 8,
        "ram_gb": 7.840827941894531,
        "memory_bandwidth_gb_s": null,
        "gpu": "NVIDIA GeForce GTX 1650",
        "vram_gb": 4,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": null,
        "gpu_count": 1,
        "gpu_org": "nvidia",
        "cpu_org": null
      },
      "sheet": null,
      "signed_runs": 3,
      "signed_runs_30d": 2,
      "builds_measured": 3,
      "best_results": 2,
      "first_signed_at": "2026-10-26T10:45:15.369Z",
      "last_signed_at": "2026-10-26T10:45:15.369Z",
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 13,
        "queued": 1867
      },
      "download_mib_per_sec": 7.3,
      "uptime_days": [],
      "uptime_pct": null,
      "best_value": {
        "model": "Sarvam-1",
        "page": "/inferences/sarvam/1",
        "build": "Q4_K_M (4-bit) by bartowski",
        "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
        "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
        "engine": "ollama",
        "tokens_per_sec": 23.48,
        "cost_usd_per_million": 7.5107,
        "measured_at": "2026-10-26T10:45:15.368Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220df04c29e6a0ad2ef0dc554f22fdbb03346428fde27cc62255d9dd9c61412954d",
          "signature": "XQOANHZUv5uZTkhIKpy3FUmfU2Fpan+cDJEruzYZSL21nNB8Pq06FL4PXreMRE3oXDqN17IOJKpdRt69sOtcBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "1nhtmpsdU4jH2I1tMjpeMpZXlxoMi9hCRX4Cq4pe8yQ="
        },
        "tokens_per_usd_hour": 37
      },
      "opened": [
        {
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "family": "sarvam",
          "member": "1",
          "at": "2026-10-26T10:45:15.369Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-26",
          "runs": 3
        }
      ],
      "runs_by_family": [
        {
          "family": "gemma-4",
          "runs": 1
        },
        {
          "family": "musicgen",
          "runs": 1
        },
        {
          "family": "sarvam",
          "runs": 1
        }
      ],
      "rooms": [
        "IN_600001",
        "IN_700108",
        "IN_751030",
        "IN_751031",
        "IN_769002"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "opener",
          "label": "First to measure 1 build"
        }
      ],
      "fastest": {
        "model": "Sarvam-1",
        "page": "/inferences/sarvam/1",
        "build": "Q4_K_M (4-bit) by bartowski",
        "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
        "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
        "engine": "ollama",
        "tokens_per_sec": 23.48,
        "cost_usd_per_million": 7.5107,
        "measured_at": "2026-10-26T10:45:15.368Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220df04c29e6a0ad2ef0dc554f22fdbb03346428fde27cc62255d9dd9c61412954d",
          "signature": "XQOANHZUv5uZTkhIKpy3FUmfU2Fpan+cDJEruzYZSL21nNB8Pq06FL4PXreMRE3oXDqN17IOJKpdRt69sOtcBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "1nhtmpsdU4jH2I1tMjpeMpZXlxoMi9hCRX4Cq4pe8yQ="
        },
        "tokens_per_usd_hour": 37
      },
      "cheapest": {
        "model": "Sarvam-1",
        "page": "/inferences/sarvam/1",
        "build": "Q4_K_M (4-bit) by bartowski",
        "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
        "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
        "engine": "ollama",
        "tokens_per_sec": 23.48,
        "cost_usd_per_million": 7.5107,
        "measured_at": "2026-10-26T10:45:15.368Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220df04c29e6a0ad2ef0dc554f22fdbb03346428fde27cc62255d9dd9c61412954d",
          "signature": "XQOANHZUv5uZTkhIKpy3FUmfU2Fpan+cDJEruzYZSL21nNB8Pq06FL4PXreMRE3oXDqN17IOJKpdRt69sOtcBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "1nhtmpsdU4jH2I1tMjpeMpZXlxoMi9hCRX4Cq4pe8yQ="
        },
        "tokens_per_usd_hour": 37
      },
      "builds": [
        {
          "model": "Sarvam-1",
          "page": "/inferences/sarvam/1",
          "build": "Q4_K_M (4-bit) by bartowski",
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama",
          "tokens_per_sec": 23.48,
          "cost_usd_per_million": 7.5107,
          "measured_at": "2026-10-26T10:45:15.368Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220df04c29e6a0ad2ef0dc554f22fdbb03346428fde27cc62255d9dd9c61412954d",
            "signature": "XQOANHZUv5uZTkhIKpy3FUmfU2Fpan+cDJEruzYZSL21nNB8Pq06FL4PXreMRE3oXDqN17IOJKpdRt69sOtcBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "1nhtmpsdU4jH2I1tMjpeMpZXlxoMi9hCRX4Cq4pe8yQ="
          },
          "tokens_per_usd_hour": 37
        },
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama",
          "tokens_per_sec": 9.95,
          "cost_usd_per_million": 17.9178,
          "measured_at": "2026-10-26T10:45:15.368Z",
          "context_tokens": 131072,
          "proof": {
            "node_did": "did:epn:002408011220df04c29e6a0ad2ef0dc554f22fdbb03346428fde27cc62255d9dd9c61412954d",
            "signature": "7ghbLIM0mYzc7+MMSlVzUcXO9oRynEgJKiNgmYUASmPyB7MLJgr762sG79s2MDqKFWMAQCKPN893YejivYy9Bw==",
            "signing_payload_b64": "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",
            "payload_sha256": "lVrR+u4k4x1+bGFTjgCnr1O6OldbXHw8jAyvI7fgT50="
          },
          "tokens_per_usd_hour": 16
        }
      ],
      "rank": 24,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": null,
        "cpu_all_cores": null,
        "cpu_one_core": null,
        "disk_read": null,
        "download": 30
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "15d9e9",
      "did": "did:epn:0024080112209a6e5031554e37f7858a302801b5f46aec5acc4e85dfbda2caada0cd9315d9e9",
      "name": "Gram ·15d9e9",
      "claimed": false,
      "online": false,
      "last_seen": null,
      "version": null,
      "region": null,
      "hardware": {
        "os": null,
        "arch": null,
        "class": null,
        "cpu_model": null,
        "cpu_clock_ghz": null,
        "cpus": null,
        "ram_gb": null,
        "memory_bandwidth_gb_s": null,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": null,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": null
      },
      "sheet": null,
      "signed_runs": 3,
      "signed_runs_30d": 0,
      "builds_measured": 3,
      "best_results": 0,
      "first_signed_at": null,
      "last_signed_at": null,
      "awake_pct": null,
      "queue": null,
      "download_mib_per_sec": null,
      "uptime_days": [],
      "uptime_pct": null,
      "best_value": null,
      "opened": [],
      "runs_by_day": [],
      "runs_by_family": [
        {
          "family": "other",
          "runs": 3
        }
      ],
      "rooms": [],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 25,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": null,
        "cpu_all_cores": null,
        "cpu_one_core": null,
        "disk_read": null,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "20d457",
      "did": "did:epn:00240801122042cc1907716d8ae975de47325f756b134b10d1f2844490761f31682e6520d457",
      "name": "Gram ·20d457",
      "claimed": true,
      "online": false,
      "last_seen": "2026-10-30T00:02:04.998Z",
      "version": "d2622d3",
      "region": {
        "code": "BE_1930",
        "city": "Zaventem",
        "state": "Vlaams-Brabant"
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2.2,
        "cpus": 4,
        "ram_gb": 31.346603393554688,
        "memory_bandwidth_gb_s": 1.3,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 4021.2,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
          "clock_ghz": 2.2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 256,
          "l3_mb": 55,
          "single_core_hashes_per_sec": 129.6,
          "multi_core_hashes_per_sec": 408,
          "sustained_ratio": 1.5,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-10-29T23:15:08.700Z",
          "measured_in": "BE_1930",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 1.3,
          "latency_ns": 55,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-10-29T23:15:09.395Z",
          "measured_in": "BE_1930",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 4021.2,
          "write_mb_s": 1.5,
          "random_read_iops": 400019,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-10-29T23:15:52.985Z",
          "measured_in": "BE_1930",
          "signed": true
        }
      },
      "signed_runs": 1,
      "signed_runs_30d": 1,
      "builds_measured": 1,
      "best_results": 0,
      "first_signed_at": "2026-10-29T23:29:55.076Z",
      "last_signed_at": "2026-10-29T23:59:55.077Z",
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 0,
        "queued": 233
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-29",
          "hours": 1
        },
        {
          "date": "2026-10-30",
          "hours": 1
        }
      ],
      "uptime_pct": 4.2,
      "best_value": null,
      "opened": [],
      "runs_by_day": [
        {
          "date": "2026-10-29",
          "runs": 3
        }
      ],
      "runs_by_family": [
        {
          "family": "glm-4-7",
          "runs": 1
        }
      ],
      "rooms": [],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 26,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 30,
        "cpu_all_cores": 29,
        "cpu_one_core": 18,
        "disk_read": 71,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "7e05a1",
      "did": "did:epn:0024080112202d032ebbd8f57e5e557f0d78039f1846a3c3b9c0137c13acf1490370787e05a1",
      "name": "Gram ·7e05a1",
      "claimed": true,
      "online": false,
      "last_seen": "2026-10-30T00:00:10.556Z",
      "version": "d2622d3",
      "region": {
        "code": "US_20511",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2.2,
        "cpus": 4,
        "ram_gb": 31.346603393554688,
        "memory_bandwidth_gb_s": 4.7,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 1205.3,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
          "clock_ghz": 2.2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 256,
          "l3_mb": 55,
          "single_core_hashes_per_sec": 119.7,
          "multi_core_hashes_per_sec": 260.6,
          "sustained_ratio": 1.4,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-10-29T23:13:29.334Z",
          "measured_in": "US_20511",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 4.7,
          "latency_ns": 73.3,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-10-29T23:13:30.044Z",
          "measured_in": "US_20511",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 1205.3,
          "write_mb_s": 3.3,
          "random_read_iops": 320888,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-10-29T23:13:49.488Z",
          "measured_in": "US_20511",
          "signed": true
        }
      },
      "signed_runs": 1,
      "signed_runs_30d": 1,
      "builds_measured": 1,
      "best_results": 0,
      "first_signed_at": "2026-10-29T23:27:55.931Z",
      "last_signed_at": "2026-10-29T23:42:55.933Z",
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 0,
        "queued": 233
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-29",
          "hours": 1
        },
        {
          "date": "2026-10-30",
          "hours": 1
        }
      ],
      "uptime_pct": 4.2,
      "best_value": null,
      "opened": [],
      "runs_by_day": [
        {
          "date": "2026-10-29",
          "runs": 2
        }
      ],
      "runs_by_family": [
        {
          "family": "glm-4-7",
          "runs": 1
        }
      ],
      "rooms": [],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 27,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 74,
        "cpu_all_cores": 18,
        "cpu_one_core": 14,
        "disk_read": 14,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "0da967",
      "did": "did:epn:002408011220cd60982a5c9fe1868084dcb781c14a37442d3f66d60505960b3bfe880e0da967",
      "name": "Gram ·0da967",
      "claimed": true,
      "online": false,
      "last_seen": "2026-10-29T23:14:05.459Z",
      "version": "21cdd90",
      "region": {
        "code": "BE_1930",
        "city": "Zaventem",
        "state": "Vlaams-Brabant"
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2.2,
        "cpus": 4,
        "ram_gb": 31.346603393554688,
        "memory_bandwidth_gb_s": 1.8,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 3967.7,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
          "clock_ghz": 2.2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 256,
          "l3_mb": 55,
          "single_core_hashes_per_sec": 201.9,
          "multi_core_hashes_per_sec": 508.6,
          "sustained_ratio": 1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-10-29T21:15:10.052Z",
          "measured_in": "BE_1930",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 1.8,
          "latency_ns": 90.2,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-10-29T21:15:10.785Z",
          "measured_in": "BE_1930",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 3967.7,
          "write_mb_s": 14.2,
          "random_read_iops": 245829,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-10-29T21:15:15.536Z",
          "measured_in": "BE_1930",
          "signed": true
        }
      },
      "signed_runs": 1,
      "signed_runs_30d": 1,
      "builds_measured": 1,
      "best_results": 0,
      "first_signed_at": "2026-10-29T21:44:56.487Z",
      "last_signed_at": "2026-10-29T22:59:56.487Z",
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 0,
        "queued": 1377
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-29",
          "hours": 3
        }
      ],
      "uptime_pct": 12.5,
      "best_value": null,
      "opened": [],
      "runs_by_day": [
        {
          "date": "2026-10-29",
          "runs": 6
        }
      ],
      "runs_by_family": [
        {
          "family": "glm-4-7",
          "runs": 1
        }
      ],
      "rooms": [
        "BE_1930"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 28,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 44,
        "cpu_all_cores": 50,
        "cpu_one_core": 50,
        "disk_read": 68,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "75e1cc",
      "did": "did:epn:002408011220b18403cfb36af164c5b1faf2d59cbf6a61d8992f44110697a615a76ee975e1cc",
      "name": "Gram ·75e1cc",
      "claimed": true,
      "online": false,
      "last_seen": "2026-10-29T23:11:56.098Z",
      "version": "21cdd90",
      "region": {
        "code": "US_51503",
        "city": "Council Bluffs",
        "state": "Iowa"
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2.2,
        "cpus": 4,
        "ram_gb": 31.346599578857422,
        "memory_bandwidth_gb_s": 1.7,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 3480.6,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
          "clock_ghz": 2.2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 256,
          "l3_mb": 55,
          "single_core_hashes_per_sec": 173.3,
          "multi_core_hashes_per_sec": 480.2,
          "sustained_ratio": 1.1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-10-29T20:03:55.568Z",
          "measured_in": "US_51503",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 1.7,
          "latency_ns": 29.5,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-10-29T20:03:56.210Z",
          "measured_in": "US_51503",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 3480.6,
          "write_mb_s": 228.3,
          "random_read_iops": 351402,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-10-29T20:03:56.739Z",
          "measured_in": "US_51503",
          "signed": true
        }
      },
      "signed_runs": 1,
      "signed_runs_30d": 1,
      "builds_measured": 1,
      "best_results": 0,
      "first_signed_at": "2026-10-29T20:23:35.578Z",
      "last_signed_at": "2026-10-29T22:58:35.581Z",
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 0,
        "queued": 1377
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-29",
          "hours": 4
        }
      ],
      "uptime_pct": 16.7,
      "best_value": null,
      "opened": [],
      "runs_by_day": [
        {
          "date": "2026-10-29",
          "runs": 11
        }
      ],
      "runs_by_family": [
        {
          "family": "glm-4-7",
          "runs": 1
        }
      ],
      "rooms": [
        "US_51503"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 29,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 37,
        "cpu_all_cores": 46,
        "cpu_one_core": 29,
        "disk_read": 57,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "d83701",
      "did": "did:epn:0024080112203cd1a47e5f3a4c215a8015d60173cafda7071fe7b015a9f9de94b6218ed83701",
      "name": "Gram ·d83701",
      "claimed": false,
      "online": false,
      "last_seen": "2026-10-29T21:13:59.894Z",
      "version": "21cdd90",
      "region": {
        "code": "AU_2001",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "AMD EPYC 7B12",
        "cpu_clock_ghz": null,
        "cpus": 4,
        "ram_gb": 31.346603393554688,
        "memory_bandwidth_gb_s": 8.9,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 8327,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "amd"
      },
      "sheet": {
        "cpu": {
          "model": "AMD EPYC 7B12",
          "model_raw": "AMD EPYC 7B12",
          "clock_ghz": null,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 512,
          "l3_mb": 16,
          "single_core_hashes_per_sec": 1403.2,
          "multi_core_hashes_per_sec": 4611.1,
          "sustained_ratio": 1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-10-29T19:41:32.618Z",
          "measured_in": "AU_2001",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 8.9,
          "latency_ns": 28.6,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-10-29T19:41:33.246Z",
          "measured_in": "AU_2001",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 8327,
          "write_mb_s": 237.4,
          "random_read_iops": 478941,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-10-29T19:41:33.742Z",
          "measured_in": "AU_2001",
          "signed": true
        }
      },
      "signed_runs": 1,
      "signed_runs_30d": 1,
      "builds_measured": 1,
      "best_results": 0,
      "first_signed_at": "2026-10-29T19:55:48.922Z",
      "last_signed_at": "2026-10-29T21:10:48.915Z",
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 0,
        "queued": 1377
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-29",
          "hours": 3
        }
      ],
      "uptime_pct": 12.5,
      "best_value": null,
      "opened": [],
      "runs_by_day": [
        {
          "date": "2026-10-29",
          "runs": 6
        }
      ],
      "runs_by_family": [
        {
          "family": "glm-4-7",
          "runs": 1
        }
      ],
      "rooms": [
        "AU_2001"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 30,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 85,
        "cpu_all_cores": 79,
        "cpu_one_core": 79,
        "disk_read": 93,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "5ac377",
      "did": "did:epn:002408011220bade96bfeca8ed3571999be84d2796bcc42fc7cc8c677f2b64e2b1b4885ac377",
      "name": "Gram ·5ac377",
      "claimed": false,
      "online": false,
      "last_seen": "2026-10-29T20:02:24.867Z",
      "version": "21cdd90",
      "region": {
        "code": "AU_2001",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2.2,
        "cpus": 4,
        "ram_gb": 31.346603393554688,
        "memory_bandwidth_gb_s": 0.4,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 3874.5,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
          "clock_ghz": 2.2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 256,
          "l3_mb": 55,
          "single_core_hashes_per_sec": 186,
          "multi_core_hashes_per_sec": 364.3,
          "sustained_ratio": 0.8,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-10-29T19:39:45.125Z",
          "measured_in": "AU_2001",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 0.4,
          "latency_ns": 191.6,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-10-29T19:39:46.062Z",
          "measured_in": "AU_2001",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 3874.5,
          "write_mb_s": 31.2,
          "random_read_iops": 292012,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-10-29T19:39:48.381Z",
          "measured_in": "AU_2001",
          "signed": true
        }
      },
      "signed_runs": 1,
      "signed_runs_30d": 1,
      "builds_measured": 1,
      "best_results": 0,
      "first_signed_at": "2026-10-29T19:54:05.041Z",
      "last_signed_at": "2026-10-29T19:54:05.041Z",
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 0,
        "queued": 440
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-29",
          "hours": 2
        }
      ],
      "uptime_pct": 8.3,
      "best_value": null,
      "opened": [
        {
          "ref": "hf.co/unsloth/GLM-4.7-Flash-GGUF:Q4_K_M",
          "family": "glm-4-7",
          "member": "flash",
          "at": "2026-10-29T19:54:05.041Z"
        }
      ],
      "runs_by_day": [
        {
          "date": "2026-10-29",
          "runs": 1
        }
      ],
      "runs_by_family": [
        {
          "family": "glm-4-7",
          "runs": 1
        }
      ],
      "rooms": [],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "opener",
          "label": "First to measure 1 build"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 31,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 7,
        "cpu_all_cores": 21,
        "cpu_one_core": 46,
        "disk_read": 64,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "73d610",
      "did": "did:epn:0024080112206b9b2bda8b5991a192031c655404858504f027bf9b3324438bbd4365cc73d610",
      "name": "Gram ·73d610",
      "claimed": false,
      "online": false,
      "last_seen": "2026-10-29T19:39:03.420Z",
      "version": "fafc916",
      "region": {
        "code": "AU_2001",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2.2,
        "cpus": 4,
        "ram_gb": 31.346603393554688,
        "memory_bandwidth_gb_s": 2.1,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 4285.5,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
          "clock_ghz": 2.2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 256,
          "l3_mb": 55,
          "single_core_hashes_per_sec": 279.1,
          "multi_core_hashes_per_sec": 594.3,
          "sustained_ratio": 1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-10-29T18:25:25.424Z",
          "measured_in": "AU_2001",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 2.1,
          "latency_ns": 97.5,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-10-29T18:25:26.130Z",
          "measured_in": "AU_2001",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 4285.5,
          "write_mb_s": 222.4,
          "random_read_iops": 488334,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-10-29T18:25:26.665Z",
          "measured_in": "AU_2001",
          "signed": true
        }
      },
      "signed_runs": 1,
      "signed_runs_30d": 1,
      "builds_measured": 1,
      "best_results": 1,
      "first_signed_at": "2026-10-29T18:40:04.810Z",
      "last_signed_at": "2026-10-29T19:25:04.813Z",
      "awake_pct": 100,
      "queue": {
        "measured": 1,
        "failed": 0,
        "queued": 1377
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-29",
          "hours": 2
        }
      ],
      "uptime_pct": 8.3,
      "best_value": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": null,
        "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
        "ref": "llama3.2:1b",
        "engine": "ollama-host",
        "tokens_per_sec": 10.79,
        "cost_usd_per_million": 2.8196,
        "measured_at": "2026-11-14T20:22:46.948Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
          "signature": "YnmWexpPC8JA+V6mE8GoELi444EXbvlieEu8EMkIs3sIi9Ru/YJxvPVDAp2KeeN7xyGH0bGJWR8ATiOhrJ5rBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "9/LgoM2fHYoSl4S+4c/CUnKMNv0+w53a/nJDIz2a6cE="
        },
        "tokens_per_usd_hour": 99
      },
      "opened": [],
      "runs_by_day": [
        {
          "date": "2026-10-29",
          "runs": 3
        }
      ],
      "runs_by_family": [
        {
          "family": "llama-3-2",
          "runs": 1
        }
      ],
      "rooms": [],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": null,
        "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
        "ref": "llama3.2:1b",
        "engine": "ollama-host",
        "tokens_per_sec": 10.79,
        "cost_usd_per_million": 2.8196,
        "measured_at": "2026-11-14T20:22:46.948Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
          "signature": "YnmWexpPC8JA+V6mE8GoELi444EXbvlieEu8EMkIs3sIi9Ru/YJxvPVDAp2KeeN7xyGH0bGJWR8ATiOhrJ5rBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "9/LgoM2fHYoSl4S+4c/CUnKMNv0+w53a/nJDIz2a6cE="
        },
        "tokens_per_usd_hour": 99
      },
      "cheapest": {
        "model": "Llama 3.2 1B",
        "page": "/inferences/llama-3-2/1b",
        "build": null,
        "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
        "ref": "llama3.2:1b",
        "engine": "ollama-host",
        "tokens_per_sec": 10.79,
        "cost_usd_per_million": 2.8196,
        "measured_at": "2026-11-14T20:22:46.948Z",
        "context_tokens": 8192,
        "proof": {
          "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
          "signature": "YnmWexpPC8JA+V6mE8GoELi444EXbvlieEu8EMkIs3sIi9Ru/YJxvPVDAp2KeeN7xyGH0bGJWR8ATiOhrJ5rBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "9/LgoM2fHYoSl4S+4c/CUnKMNv0+w53a/nJDIz2a6cE="
        },
        "tokens_per_usd_hour": 99
      },
      "builds": [
        {
          "model": "Llama 3.2 1B",
          "page": "/inferences/llama-3-2/1b",
          "build": null,
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "tokens_per_sec": 10.79,
          "cost_usd_per_million": 2.8196,
          "measured_at": "2026-11-14T20:22:46.948Z",
          "context_tokens": 8192,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "YnmWexpPC8JA+V6mE8GoELi444EXbvlieEu8EMkIs3sIi9Ru/YJxvPVDAp2KeeN7xyGH0bGJWR8ATiOhrJ5rBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "9/LgoM2fHYoSl4S+4c/CUnKMNv0+w53a/nJDIz2a6cE="
          },
          "tokens_per_usd_hour": 99
        }
      ],
      "rank": 32,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 48,
        "cpu_all_cores": 57,
        "cpu_one_core": 57,
        "disk_read": 79,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "748763",
      "did": "did:epn:002408011220f13b9ea7c0bdcd07317be12c9715d458943a2ff66267447afeb54bee84748763",
      "name": "Gram ·748763",
      "claimed": false,
      "online": false,
      "last_seen": "2026-10-29T19:36:55.779Z",
      "version": "fafc916",
      "region": {
        "code": "US_97058",
        "city": "The Dalles",
        "state": "Oregon"
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/32g",
        "cpu_model": "Intel Xeon",
        "cpu_clock_ghz": 2.2,
        "cpus": 4,
        "ram_gb": 31.346603393554688,
        "memory_bandwidth_gb_s": 1.5,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 3662.8,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "intel"
      },
      "sheet": {
        "cpu": {
          "model": "Intel Xeon",
          "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
          "clock_ghz": 2.2,
          "cores_logical": 4,
          "cores_physical": 2,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 256,
          "l3_mb": 55,
          "single_core_hashes_per_sec": 218.7,
          "multi_core_hashes_per_sec": 446,
          "sustained_ratio": 1,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-10-29T18:25:16.143Z",
          "measured_in": "US_97058",
          "signed": true
        },
        "memory": {
          "total_gb": 31.3,
          "bandwidth_gb_s": 1.5,
          "latency_ns": 96.1,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-10-29T18:25:16.883Z",
          "measured_in": "US_97058",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 3662.8,
          "write_mb_s": 209.6,
          "random_read_iops": 286338,
          "fs_type": "overlayfs",
          "device_model": null,
          "measured_at": "2026-10-29T18:25:19.199Z",
          "measured_in": "US_97058",
          "signed": true
        }
      },
      "signed_runs": 1,
      "signed_runs_30d": 1,
      "builds_measured": 1,
      "best_results": 0,
      "first_signed_at": "2026-10-29T18:39:40.090Z",
      "last_signed_at": "2026-10-29T19:24:40.096Z",
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 0,
        "queued": 1256
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-29",
          "hours": 2
        }
      ],
      "uptime_pct": 8.3,
      "best_value": null,
      "opened": [],
      "runs_by_day": [
        {
          "date": "2026-10-29",
          "runs": 4
        }
      ],
      "runs_by_family": [
        {
          "family": "llama-3-2",
          "runs": 1
        }
      ],
      "rooms": [
        "US_97058"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 33,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 33,
        "cpu_all_cores": 39,
        "cpu_one_core": 54,
        "disk_read": 61,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "e78fd2",
      "did": "did:epn:00240801122095368b040a04cc8b46563ce9ac25e091edccd520d7753bb0a0093f9490e78fd2",
      "name": "Gram ·e78fd2",
      "claimed": false,
      "online": false,
      "last_seen": "2026-10-19T11:34:49.856Z",
      "version": "9ef55aa",
      "region": null,
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/nvidia/32g",
        "cpu_model": null,
        "cpu_clock_ghz": null,
        "cpus": 4,
        "ram_gb": 31.34674072265625,
        "memory_bandwidth_gb_s": null,
        "gpu": "Tesla T4",
        "vram_gb": 15,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": null,
        "gpu_count": 1,
        "gpu_org": "nvidia",
        "cpu_org": null
      },
      "sheet": null,
      "signed_runs": 1,
      "signed_runs_30d": 1,
      "builds_measured": 1,
      "best_results": 1,
      "first_signed_at": "2026-10-19T11:28:03.450Z",
      "last_signed_at": "2026-10-19T11:28:03.450Z",
      "awake_pct": 100,
      "queue": {
        "measured": 1,
        "failed": 0,
        "queued": 139
      },
      "download_mib_per_sec": 127.9,
      "uptime_days": [],
      "uptime_pct": null,
      "best_value": {
        "model": "Gemma 4 E2B",
        "page": "/inferences/gemma-4/e2b",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
        "engine": "ollama-host",
        "tokens_per_sec": 77.58,
        "cost_usd_per_million": 1.7757,
        "measured_at": "2026-10-19T11:28:03.450Z",
        "context_tokens": 4096,
        "proof": {
          "node_did": "did:epn:00240801122095368b040a04cc8b46563ce9ac25e091edccd520d7753bb0a0093f9490e78fd2",
          "signature": "OVe3nttZDdDnZkX344L+tBDQQsFslXcX4T6fYZDm8AXTtAE2Ki6wp2w1ikvYhW/kC46HbqrTXOjoz9dpmpVQCQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "Yn8g7FG6ZJeVRPyQI1hfyHm9MwQeGWZQjYqWcxW1z8M="
        },
        "tokens_per_usd_hour": 156
      },
      "opened": [],
      "runs_by_day": [
        {
          "date": "2026-10-19",
          "runs": 2
        }
      ],
      "runs_by_family": [
        {
          "family": "gemma-4",
          "runs": 1
        }
      ],
      "rooms": [],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        }
      ],
      "fastest": {
        "model": "Gemma 4 E2B",
        "page": "/inferences/gemma-4/e2b",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
        "engine": "ollama-host",
        "tokens_per_sec": 77.58,
        "cost_usd_per_million": 1.7757,
        "measured_at": "2026-10-19T11:28:03.450Z",
        "context_tokens": 4096,
        "proof": {
          "node_did": "did:epn:00240801122095368b040a04cc8b46563ce9ac25e091edccd520d7753bb0a0093f9490e78fd2",
          "signature": "OVe3nttZDdDnZkX344L+tBDQQsFslXcX4T6fYZDm8AXTtAE2Ki6wp2w1ikvYhW/kC46HbqrTXOjoz9dpmpVQCQ==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NzcuNTc3NTI3NDExNDA4MTgsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkyNDA5MjgzNDUwMTYxODkyLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwOTUzNjhiMDQwYTA0Y2M4YjQ2NTYzY2U5YWMyNWUwOTFlZGNjZDUyMGQ3NzUzYmIwYTAwOTNmOTQ5MGU3OGZkMiIsImV4dHJhIjp7Im1vZGVsIjoiaGYuY28vdW5zbG90aC9nZW1tYS00LUUyQi1pdC1HR1VGOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50IjozNSwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjo5Ny43NzMxNjAwMDAwMDAwMiwidG90YWxfdG9rZW5zIjo3NTg1fX0=",
          "payload_sha256": "Yn8g7FG6ZJeVRPyQI1hfyHm9MwQeGWZQjYqWcxW1z8M="
        },
        "tokens_per_usd_hour": 156
      },
      "cheapest": {
        "model": "Gemma 4 E2B",
        "page": "/inferences/gemma-4/e2b",
        "build": "Q4_K_M (4-bit) by unsloth",
        "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
        "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
        "engine": "ollama-host",
        "tokens_per_sec": 77.58,
        "cost_usd_per_million": 1.7757,
        "measured_at": "2026-10-19T11:28:03.450Z",
        "context_tokens": 4096,
        "proof": {
          "node_did": "did:epn:00240801122095368b040a04cc8b46563ce9ac25e091edccd520d7753bb0a0093f9490e78fd2",
          "signature": "OVe3nttZDdDnZkX344L+tBDQQsFslXcX4T6fYZDm8AXTtAE2Ki6wp2w1ikvYhW/kC46HbqrTXOjoz9dpmpVQCQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "Yn8g7FG6ZJeVRPyQI1hfyHm9MwQeGWZQjYqWcxW1z8M="
        },
        "tokens_per_usd_hour": 156
      },
      "builds": [
        {
          "model": "Gemma 4 E2B",
          "page": "/inferences/gemma-4/e2b",
          "build": "Q4_K_M (4-bit) by unsloth",
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "tokens_per_sec": 77.58,
          "cost_usd_per_million": 1.7757,
          "measured_at": "2026-10-19T11:28:03.450Z",
          "context_tokens": 4096,
          "proof": {
            "node_did": "did:epn:00240801122095368b040a04cc8b46563ce9ac25e091edccd520d7753bb0a0093f9490e78fd2",
            "signature": "OVe3nttZDdDnZkX344L+tBDQQsFslXcX4T6fYZDm8AXTtAE2Ki6wp2w1ikvYhW/kC46HbqrTXOjoz9dpmpVQCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "Yn8g7FG6ZJeVRPyQI1hfyHm9MwQeGWZQjYqWcxW1z8M="
          },
          "tokens_per_usd_hour": 156
        }
      ],
      "rank": 34,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": null,
        "cpu_all_cores": null,
        "cpu_one_core": null,
        "disk_read": null,
        "download": 90
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "a5b1da",
      "did": "did:epn:00240801122005b4f7d449d674902e2d9bba7985520460b5decea661244604d4debf80a5b1da",
      "name": "Gram ·a5b1da",
      "claimed": false,
      "online": false,
      "last_seen": "2026-11-14T09:51:20.874Z",
      "version": "afa811d",
      "region": {
        "code": "AM_0002",
        "city": null,
        "state": null
      },
      "hardware": {
        "os": "darwin",
        "arch": "arm64",
        "class": "darwin/arm64/apple/16g",
        "cpu_model": "Apple M1",
        "cpu_clock_ghz": null,
        "cpus": 8,
        "ram_gb": 16,
        "memory_bandwidth_gb_s": 27.7,
        "gpu": "Apple M1",
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 6263.3,
        "gpu_count": 1,
        "gpu_org": "apple",
        "cpu_org": "apple"
      },
      "sheet": {
        "cpu": {
          "model": "Apple M1",
          "model_raw": "Apple M1",
          "clock_ghz": null,
          "cores_logical": 8,
          "cores_physical": 8,
          "threads_per_core": 1,
          "sockets": 1,
          "l2_kb": 4096,
          "l3_mb": null,
          "single_core_hashes_per_sec": 2002,
          "multi_core_hashes_per_sec": 13965.2,
          "sustained_ratio": 1.2,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-14T08:03:47.425Z",
          "measured_in": "AM_0002",
          "signed": true
        },
        "memory": {
          "total_gb": 16,
          "bandwidth_gb_s": 27.7,
          "latency_ns": 7,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-14T08:03:47.969Z",
          "measured_in": "AM_0002",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 6263.3,
          "write_mb_s": 1266.5,
          "random_read_iops": 1295705,
          "fs_type": null,
          "device_model": null,
          "measured_at": "2026-11-14T08:03:48.249Z",
          "measured_in": "AM_0002",
          "signed": true
        }
      },
      "signed_runs": 1,
      "signed_runs_30d": 0,
      "builds_measured": 1,
      "best_results": 0,
      "first_signed_at": null,
      "last_signed_at": null,
      "awake_pct": 15.7,
      "queue": {
        "measured": 0,
        "failed": 2,
        "queued": 1367
      },
      "download_mib_per_sec": 20.2,
      "uptime_days": [
        {
          "date": "2026-11-04",
          "hours": 3
        },
        {
          "date": "2026-11-05",
          "hours": 18
        },
        {
          "date": "2026-11-06",
          "hours": 17
        },
        {
          "date": "2026-11-07",
          "hours": 14
        },
        {
          "date": "2026-11-08",
          "hours": 16
        },
        {
          "date": "2026-11-09",
          "hours": 11
        },
        {
          "date": "2026-11-12",
          "hours": 16
        },
        {
          "date": "2026-11-13",
          "hours": 18
        },
        {
          "date": "2026-11-14",
          "hours": 10
        }
      ],
      "uptime_pct": 56.9,
      "best_value": null,
      "opened": [],
      "runs_by_day": [],
      "runs_by_family": [
        {
          "family": "whisper",
          "runs": 1
        }
      ],
      "rooms": [
        "AM_0002",
        "DG1h70GzVONqlxjluvnM",
        "ecfmNiLz2x7vWNymNbtq",
        "EUTOPIA_FOUNDATION",
        "j9p5QqSCECG55k3Qf98Z",
        "rSqKUZuufpk7scsG9rGP"
      ],
      "awards": [
        {
          "id": "runs-1",
          "label": "First signed result"
        },
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 35,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": 96,
        "cpu_all_cores": 89,
        "cpu_one_core": 93,
        "disk_read": 89,
        "download": 40
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    },
    {
      "id": "c52689",
      "did": "did:epn:0024080112208bedb76b4021960d095907e369b1fc64d3a408d9a95f69870e6691c746c52689",
      "name": "Gram ·c52689",
      "claimed": true,
      "online": true,
      "last_seen": "2026-11-16T00:22:53.133Z",
      "version": "dbd417d",
      "region": {
        "code": "IN_500032",
        "city": "Hyderabad",
        "state": "Telangana"
      },
      "hardware": {
        "os": "linux",
        "arch": "x86_64",
        "class": "linux/amd64/none/4g",
        "cpu_model": "AMD EPYC 7551",
        "cpu_clock_ghz": null,
        "cpus": 2,
        "ram_gb": 0.9316482543945312,
        "memory_bandwidth_gb_s": null,
        "gpu": null,
        "vram_gb": null,
        "gpu_tflops_fp16": null,
        "disk_read_mb_s": 25.8,
        "gpu_count": 0,
        "gpu_org": null,
        "cpu_org": "amd"
      },
      "sheet": {
        "cpu": {
          "model": "AMD EPYC 7551",
          "model_raw": "AMD EPYC 7551 32-Core Processor",
          "clock_ghz": null,
          "cores_logical": 2,
          "cores_physical": 1,
          "threads_per_core": 2,
          "sockets": 1,
          "l2_kb": 512,
          "l3_mb": 16,
          "single_core_hashes_per_sec": 14.3,
          "multi_core_hashes_per_sec": 4.6,
          "sustained_ratio": 0.5,
          "method": "sha256(1MB) single-threaded",
          "measured_at": "2026-11-15T23:55:14.931Z",
          "measured_in": "IN_500032",
          "signed": true
        },
        "memory": {
          "total_gb": 0.9,
          "bandwidth_gb_s": 0,
          "latency_ns": 2200.8,
          "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
          "measured_at": "2026-11-15T20:22:42.038Z",
          "measured_in": "IN_500032",
          "signed": true
        },
        "gpu": null,
        "storage": {
          "read_mb_s": 25.8,
          "write_mb_s": 11.2,
          "random_read_iops": 44,
          "fs_type": "ext4",
          "device_model": "BlockVolume",
          "measured_at": "2026-11-15T20:22:55.642Z",
          "measured_in": "IN_500032",
          "signed": true
        }
      },
      "signed_runs": 0,
      "signed_runs_30d": 0,
      "builds_measured": 0,
      "best_results": 0,
      "first_signed_at": null,
      "last_signed_at": null,
      "awake_pct": 100,
      "queue": {
        "measured": 0,
        "failed": 0,
        "queued": 1408
      },
      "download_mib_per_sec": null,
      "uptime_days": [
        {
          "date": "2026-10-30",
          "hours": 5
        },
        {
          "date": "2026-10-31",
          "hours": 24
        },
        {
          "date": "2026-11-01",
          "hours": 24
        },
        {
          "date": "2026-11-02",
          "hours": 24
        },
        {
          "date": "2026-11-03",
          "hours": 24
        },
        {
          "date": "2026-11-04",
          "hours": 24
        },
        {
          "date": "2026-11-05",
          "hours": 24
        },
        {
          "date": "2026-11-06",
          "hours": 18
        },
        {
          "date": "2026-11-07",
          "hours": 11
        },
        {
          "date": "2026-11-08",
          "hours": 13
        },
        {
          "date": "2026-11-09",
          "hours": 19
        },
        {
          "date": "2026-11-10",
          "hours": 7
        },
        {
          "date": "2026-11-11",
          "hours": 12
        },
        {
          "date": "2026-11-12",
          "hours": 19
        },
        {
          "date": "2026-11-13",
          "hours": 22
        },
        {
          "date": "2026-11-14",
          "hours": 22
        },
        {
          "date": "2026-11-15",
          "hours": 21
        },
        {
          "date": "2026-11-16",
          "hours": 1
        }
      ],
      "uptime_pct": 72.7,
      "best_value": null,
      "opened": [],
      "runs_by_day": [],
      "runs_by_family": [],
      "rooms": [
        "0oaKlVqk2eNkyh25H78Z",
        "26NRJ3lXsSmqiI4k6N4X",
        "asPepgKbRT1dsKNkRGZd",
        "IN_500032"
      ],
      "awards": [
        {
          "id": "sheet",
          "label": "Machine benchmarked and signed"
        }
      ],
      "fastest": null,
      "cheapest": null,
      "builds": [],
      "rank": 36,
      "of": 36,
      "percentiles": {
        "gpu_tflops": null,
        "memory_bandwidth": null,
        "cpu_all_cores": 0,
        "cpu_one_core": 4,
        "disk_read": 0,
        "download": null
      },
      "medians": {
        "gpu_tflops": 52.7,
        "memory_bandwidth": 2.4,
        "cpu_all_cores": 508.6,
        "cpu_one_core": 201.9,
        "disk_read": 2983.7,
        "download": 54.3
      }
    }
  ],
  "logos": {
    "0xSero": "https://cdn-avatars.huggingface.co/v1/production/uploads/68e3c4dbcf2503efff7eb64f/fWaG6wdfpCoiIkPuBnhhy.png",
    "3MPER0RR": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/CsOkxydAt-MTdCcELsYrD.png",
    "42ailab": "https://www.gravatar.com/avatar/4a24b0263f4cf4e7a478624f812179ff?d=retro&size=100",
    "AEON-7": "https://cdn-avatars.huggingface.co/v1/production/uploads/69d6c57f86d2016d8ffb8e73/sdMn1F74M7w9dj-tYzr0D.jpeg",
    "AIArchiveInfo": "https://www.gravatar.com/avatar/9eac872bb43bf3b9709cfda2909fa63a?d=retro&size=100",
    "Abiray": "https://cdn-avatars.huggingface.co/v1/production/uploads/672622efdf9cce918fdc713f/2wilqSZUrc1H5aGXV54Pp.png",
    "AlexGS74": "https://huggingface.co/avatars/0be2be5e1ad521bfc9f786d4e44c64dc.svg",
    "AnkitAI": "https://cdn-avatars.huggingface.co/v1/production/uploads/667bccf3d24677e9afc524f8/xdgKUKzrdeFsHb1Mb3wog.jpeg",
    "ArchiveStudio": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/uhzk6kX08hbWnOMHNZlFP.png",
    "Ares-Realm-Studios": "https://cdn-avatars.huggingface.co/v1/production/uploads/68f1f0d588e256c4ab4dcc99/Hz1YlSVXYioKqr3DlQx8H.png",
    "ArliAI": "https://cdn-avatars.huggingface.co/v1/production/uploads/6625f4a8a8d1362ebcc3851a/47WrP6UrAJPBboYMnA36j.png",
    "Austriani": "https://cdn-avatars.huggingface.co/v1/production/uploads/67781392f67a2e23ac9fd462/Fb5RnlWNJUbYSUdJG2NJx.png",
    "AyoubChLin": "https://cdn-avatars.huggingface.co/v1/production/uploads/1674223160074-63c58e4dd5a5cd2043e2f796.jpeg",
    "Beh1222": "https://huggingface.co/avatars/4b4e596e0d2561683a408713a8963ed4.svg",
    "Bercraft": "https://huggingface.co/avatars/fdbf653dbf9c14e1f69781243c204497.svg",
    "Blackfrost-AI": "https://cdn-avatars.huggingface.co/v1/production/uploads/69a27f2d114e4ac9de4dafc7/wAMHv3bx31qDjNhz0ygkV.jpeg",
    "Boopathy2861": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/-VbnMMNsI7olrZ6jw91US.png",
    "BrainForge": "https://cdn-avatars.huggingface.co/v1/production/uploads/627aac113040e646e80c55bf/pP4XXQg40Jtii7eO9-Al5.png",
    "BytedTsinghua-SIA": "https://cdn-avatars.huggingface.co/v1/production/uploads/6341375be5071b0c5cc946a3/GSHXRdrZqDOzqHrfcZ0d5.png",
    "CaaretAI": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/dQk2HF1Yhkmj5jYE5knuw.png",
    "Caldalis": "https://cdn-avatars.huggingface.co/v1/production/uploads/6a98deee9fa58bce5a2fcd61/xaOsFnRBSBPSeNT9G33cY.png",
    "CoNDeNse-AI": "https://cdn-avatars.huggingface.co/v1/production/uploads/69c61ec4faed05be5d8701f4/QO3puDUyvKCSaeBAbjPtb.png",
    "CohereLabs": "https://cdn-avatars.huggingface.co/v1/production/uploads/1678549441248-5e70f6048ce3c604d78fe133.png",
    "Darylgeero": "https://huggingface.co/avatars/24296b69e3199d9e76ae9fe684dd5c76.svg",
    "DavidAU": "https://cdn-avatars.huggingface.co/v1/production/uploads/65ea44635b64331c067d3751/yCim-7c3tm67o5wWP_6cE.jpeg",
    "Dingdust": "https://cdn-avatars.huggingface.co/v1/production/uploads/6655ca7d258f5fad31cfb170/lG9ndmNyyDq7HQ_UkDJLQ.jpeg",
    "Disty0": "https://cdn-avatars.huggingface.co/v1/production/uploads/6456af6195082f722d178522/ovlj-VCeTAawt9omkkZIz.png",
    "DreamFast": "https://cdn-avatars.huggingface.co/v1/production/uploads/62fb5a2cfcce44435d7e72ed/2nImOrUOmdij_VPMgIBEQ.png",
    "Efficient-Large-Model": "https://cdn-avatars.huggingface.co/v1/production/uploads/62b4b5beb25cb80fcf278354/SIddx3hYu5rWXA4-O3Oaj.jpeg",
    "EiadEssam": "https://huggingface.co/avatars/1947b0a806082a83f7b92648860a1063.svg",
    "EnlistedGhost": "https://cdn-avatars.huggingface.co/v1/production/uploads/683426712973b9f8f54f05aa/eIIS5E0eWc2RzBMvwClnZ.png",
    "Ex0bit": "https://cdn-avatars.huggingface.co/v1/production/uploads/1672344017443-noauth.jpeg",
    "FastVideo": "https://cdn-avatars.huggingface.co/v1/production/uploads/63565cc56d7fcf1bedb7d347/g6wWXLryVUwXcaKTuRerh.png",
    "FaustSS": "https://huggingface.co/avatars/268f1104f9c4195652216b4cb520c5f6.svg",
    "FenomAI": "https://huggingface.co/avatars/4f422a24ebaed891c322612c5b964fd2.svg",
    "FriendliAI": "https://cdn-avatars.huggingface.co/v1/production/uploads/638832b61901766b8807aeac/UOpqhkrkFUKM7HoQI_-9X.png",
    "FriskyFennec": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/p1o_EX0UKZn2_O_GHMGJ3.jpeg",
    "Fu01978": "https://huggingface.co/avatars/acec42aecdf0c59de23b6e2fefb42664.svg",
    "GnLOLot": "https://cdn-avatars.huggingface.co/v1/production/uploads/697ef6e51b9661d3b60cf978/t1huf__vktKREoT12NVzP.jpeg",
    "Guilherme34": "https://cdn-avatars.huggingface.co/v1/production/uploads/638fd4be2ddd69e70b8cd31c/Rf4KhB2c6N7Qj63MIJFwX.png",
    "Hakemz": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/vFiySCxhgGaYklvwVCvLY.png",
    "HauhauCS": "https://huggingface.co/avatars/1a732618ac3a62f81515f76145c902a6.svg",
    "HuggingFaceTB": "https://cdn-avatars.huggingface.co/v1/production/uploads/651e96991b97c9f33d26bde6/e4VK7uW5sTeCYupD0s_ob.png",
    "IPostYellow": "https://huggingface.co/avatars/2a6886cd550b02c3828bd6661e3c2ae0.svg",
    "ISTA-DASLab": "https://cdn-avatars.huggingface.co/v1/production/uploads/628e0ce4e53bbd334577fcb0/TRPtgtSavYjDJOK3S1I8M.png",
    "IbrahimSeven": "https://huggingface.co/avatars/b50717d4d6b051c97ff27d956c1ee47b.svg",
    "IlyasMoutawwakil": "https://cdn-avatars.huggingface.co/v1/production/uploads/1642598610696-noauth.jpeg",
    "Inferact": "https://cdn-avatars.huggingface.co/v1/production/uploads/61075042abc7b48470cd233b/IQaEC-7BHhOPRMgMoKe4g.jpeg",
    "Infomaniak-AI": "https://cdn-avatars.huggingface.co/v1/production/uploads/5e0eed1ffcf41d740b699666/6Dxmoz92wPulQNmZhQvz5.png",
    "Intel": "https://cdn-avatars.huggingface.co/v1/production/uploads/1616186257611-60104afcc75e19ac1738fe70.png",
    "J-Fraudster": "https://cdn-avatars.huggingface.co/v1/production/uploads/69b9061f1bbb18cf8f6ebcba/xPuIiHFfCdTtb26LfPXXI.png",
    "JANGQ-AI": "https://cdn-avatars.huggingface.co/v1/production/uploads/69b7444b4221ec9396921547/GGIwtvR1fN9DMucOL2O_l.png",
    "JW-Hubs": "https://cdn-avatars.huggingface.co/v1/production/uploads/69b9061f1bbb18cf8f6ebcba/xPuIiHFfCdTtb26LfPXXI.png",
    "Jackrong": "https://cdn-avatars.huggingface.co/v1/production/uploads/66309bd090589b7c65950665/RcOk7ysh7nEt5YlHHzauj.jpeg",
    "Jinstudio": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/zSWSYxuai27PqeM2gvZQT.png",
    "Jundot": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/P4AE_DXqeRdnfzXp-shec.jpeg",
    "KH30": "https://huggingface.co/avatars/af0ed11d55ecf3697f8a6b5be55792d5.svg",
    "Kenvenx": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/sAWHXa_5i5wjR9IFeAjyJ.jpeg",
    "Kilinskiy": "https://huggingface.co/avatars/4062c0b1a563a27b0c0d91bf82b25649.svg",
    "Koshkasa": "https://cdn-avatars.huggingface.co/v1/production/uploads/69b438a92785e17461c50952/VDJq1qpHqxAm3biCH6zMQ.jpeg",
    "Lightricks": "https://cdn-avatars.huggingface.co/v1/production/uploads/669524bcbcd81f395e8f60f6/0ynfqKEWMh_dn3h4ff1K5.png",
    "LiquidAI": "https://cdn-avatars.huggingface.co/v1/production/uploads/61b8e2ba285851687028d395/EsTgVtnM2IqVRKgPdfqcB.png",
    "LoJexLLM": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/kCTjUK2YC8oUeCI2cwKeP.jpeg",
    "LyJonathan": "https://cdn-avatars.huggingface.co/v1/production/uploads/67c0d8a97fb36ad060a5a529/TA_YEOanQB7SLLF3xsGog.jpeg",
    "Mikefeed": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/ngdqm2OG4Xc8xGGlURrrL.png",
    "Mini-Bleyz": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/eqiDGK80JwfkJqT1MLa73.png",
    "MiniMaxAI": "https://cdn-avatars.huggingface.co/v1/production/uploads/676e38ad04af5bec20bc9faf/dUd-LsZEX0H_d4qefO_g6.jpeg",
    "MuXodious": "https://cdn-avatars.huggingface.co/v1/production/uploads/65d0f78d98b4a470bf2aca80/lSw3a-hajldtu2JnlptOr.png",
    "Mungert": "https://cdn-avatars.huggingface.co/v1/production/uploads/65edf8c933c27925395181d4/JTCVSaWrJEw2c66LvvXvx.png",
    "North-ML1": "https://cdn-avatars.huggingface.co/v1/production/uploads/68955d7785865264dfdfa451/vEvWH_AFrN3g7Fg5H62zQ.png",
    "NurseCitizenDeveloper": "https://cdn-avatars.huggingface.co/v1/production/uploads/684e990e746774239d6ff9b9/JgaCbzbx98HhzklCNMAwY.png",
    "OS-Software": "https://cdn-avatars.huggingface.co/v1/production/uploads/68ee0a18436e3e1c3bc8bf6d/57Klukud_t8kVzGDtgWj2.png",
    "OfflineBase": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/9SAHIYHqnaIYTQtsbgAC_.png",
    "Olafangensan": "https://huggingface.co/avatars/83702534a8b8e6ff23594b6fff031bb2.svg",
    "OpenKing": "https://www.gravatar.com/avatar/87588edd65370e86280279a730635dbc?d=retro&size=100",
    "OpenMOSE": "https://cdn-avatars.huggingface.co/v1/production/uploads/63121469ae8896941da562c1/L9YUUhou7wlvobn9D9vz8.png",
    "OpenYourMind": "https://huggingface.co/avatars/851e428d2d59fc261c0466c2bf42810a.svg",
    "OpensourceWTF": "https://huggingface.co/avatars/b8019421f949570f3a56056ac09327b5.svg",
    "OsGo": "https://huggingface.co/avatars/cee846b1645c3ac47b6e35fc11d4c8c0.svg",
    "OsmanKhalidSiddiqui": "https://huggingface.co/avatars/60bb9ad112766f48772d966b0c50a4fd.svg",
    "OzzyGT": "https://cdn-avatars.huggingface.co/v1/production/uploads/63df091910678851bb0cd0e0/FUXFt0C-rUFSppIAu5ZDN.png",
    "Pascalburume": "https://huggingface.co/avatars/45576810e29b72e2065e031e4d994a6f.svg",
    "PeterCxy": "https://huggingface.co/avatars/9479bc9730d3a096e173299358bd184d.svg",
    "Pinaster": "https://huggingface.co/avatars/4ba6038fb36bbb0242740d24a6d74ade.svg",
    "PrimeIntellect": "https://cdn-avatars.huggingface.co/v1/production/uploads/61e020e4a343274bb132e138/H2mcdPRWtl4iKLd-OYYBc.jpeg",
    "Pspdlwplxpepdpcldps": "https://huggingface.co/avatars/4301b9f9e0671cdb0bac49ad3b409c04.svg",
    "Qrzysztof": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/s4uUshiIZw6ASsS2l4Oe7.png",
    "QuantFactory": "https://cdn-avatars.huggingface.co/v1/production/uploads/6382255fcae34727b9cc149e/ANRA_7hWosC6_2PS2cwtg.jpeg",
    "QuantFunc": "https://cdn-avatars.huggingface.co/v1/production/uploads/696de9175264ad0d9e726926/Ubo6UWwNt1o-LeNRL1_zi.webp",
    "QuantTrio": "https://cdn-avatars.huggingface.co/v1/production/uploads/647da8591a1fcad2fdbf9059/-NAwIsNJDbLUGc96bOrpm.jpeg",
    "Qun1023": "https://huggingface.co/avatars/9360b44cb8b725f980d19aa22e7c2149.svg",
    "Qwen": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
    "RadicalNotionAI": "https://cdn-avatars.huggingface.co/v1/production/uploads/658f3269a78a277803930b5f/7KjHXbR-rMyJrhXUl0MUI.png",
    "RadixArk": "https://cdn-avatars.huggingface.co/v1/production/uploads/692f6fa22fe03c133986059c/uVcDantaDXJOwc_T93yei.png",
    "ReadyArt": "https://cdn-avatars.huggingface.co/v1/production/uploads/6759e155bc947d6070775cb9/uA_aXyXpBtK9M77jSv0aP.png",
    "RedHatAI": "https://cdn-avatars.huggingface.co/v1/production/uploads/60466e4b4f40b01b66151416/cdABRow21BL0sl1vSVTPk.png",
    "ResembleAI": "https://cdn-avatars.huggingface.co/v1/production/uploads/6000a0456a2a91af974298cf/qmxSbcVBCwUtVez918IkQ.png",
    "RockMan256": "https://huggingface.co/avatars/dea86fdc7cde1fb9247f35b2ebf93c8d.svg",
    "Runware": "https://cdn-avatars.huggingface.co/v1/production/uploads/69f0a2f575b2c222876d9cb5/tzakeX-xJ2FcBT5_YXRCr.png",
    "SWivid": "https://cdn-avatars.huggingface.co/v1/production/uploads/636916bb85fef3ca96e2ef73/waHubUG9Ck-EBID9PBnWZ.jpeg",
    "Saad4web": "https://huggingface.co/avatars/b69c676d87e09c6445ed07b4aa505be2.svg",
    "Sadatsami": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/2ckewwL7KU75Ks2pu2WZh.png",
    "Sainathhh": "https://huggingface.co/avatars/0d59d77da63cfa3ef5b1d02e398f0568.svg",
    "SauravMahalik": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/X4ZqtdiqCGWzMeQVzHgWy.jpeg",
    "Sci-fi-vy": "https://cdn-avatars.huggingface.co/v1/production/uploads/659b81a78c5c6688860cb1d5/w35fSP7kHzinIqzXHfcKv.jpeg",
    "Scrappy-Doo": "https://cdn-avatars.huggingface.co/v1/production/uploads/686fba92100f867a30dbe6d8/3UWt3Ivk8tPgoJR7jSyUh.png",
    "Seif6": "https://huggingface.co/avatars/eeece8c93812d7cd52c8333f208f8502.svg",
    "SentieAI": "https://huggingface.co/avatars/027fc077785fa3917d860e1252f2e867.svg",
    "SepehrKerachi": "https://cdn-avatars.huggingface.co/v1/production/uploads/67655ecc91084548788281e7/xMpUcSVt2YPUlpiVoFUzq.jpeg",
    "Silicone-Moss": "https://cdn-avatars.huggingface.co/v1/production/uploads/631c20949edb6d320a18f479/mB6nO2spB0hWE32pXh3r5.jpeg",
    "SonGT1": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/TViZ3K4pZG2CACHSDYIbs.png",
    "Systran": "https://cdn-avatars.huggingface.co/v1/production/uploads/64fb1880c367f7b1cacd259b/kt9lzy5bR-APj6IYM5ZYE.png",
    "THUDM": "https://cdn-avatars.huggingface.co/v1/production/uploads/63033dc4e1e7f0e03a5e1a31/0BibmRdezvN6v6d2CEtd5.png",
    "TeichAI": "https://cdn-avatars.huggingface.co/v1/production/uploads/6837935ac3b7ffe0d2559ce9/-AxyvV4wfUY8uo87kNKkK.png",
    "The-freezer": "https://www.gravatar.com/avatar/f896e04087c9c5e09ebab6379a99fac6?d=retro&size=100",
    "TitanML": "https://cdn-avatars.huggingface.co/v1/production/uploads/66281f4e8bfc90408a895226/JjTVmUOUhDrHs-jEe41ek.jpeg",
    "TrevorJS": "https://cdn-avatars.huggingface.co/v1/production/uploads/6310fc1464939fabc00b8df2/TxFAW1A2vpx7myZItFdXo.png",
    "Trkaga": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/C4Ryu58FxkbSYtJeZNZLt.jpeg",
    "UCloud-org": "https://cdn-avatars.huggingface.co/v1/production/uploads/6a41d2e7c65231a89cb18bec/Y2P-dJyZ66OjW0LPx8sCW.png",
    "Uniboshi": "https://cdn-avatars.huggingface.co/v1/production/uploads/63d961e46a67e9128dd1f01a/l8ipzAcd82qin9Pa1LfPt.webp",
    "UnstableLlama": "https://cdn-avatars.huggingface.co/v1/production/uploads/6459a9cb38d95f3427a537d6/tCr1a6hInZUnQ0rnnM5Hf.png",
    "ValiantLabs": "https://cdn-avatars.huggingface.co/v1/production/uploads/63444f2687964b331809eb55/sQl9O5qStzAing24c1xdI.jpeg",
    "Walltim3": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/DIbVAs7oeev52InUz0SX5.png",
    "Wan-AI": "https://cdn-avatars.huggingface.co/v1/production/uploads/67b610677ea7952def8b29c6/N6jQbbeaa_FcUY-wI1dgG.png",
    "XXINTUANDDUI": "https://huggingface.co/avatars/b0042f43ff64070e6dcfc260ecbd67af.svg",
    "XiaomiMiMo": "https://cdn-avatars.huggingface.co/v1/production/uploads/680cb7d1233834890a64acee/5w_4aLfF-7MAyaIPOV498.jpeg",
    "Xiejiehang": "https://huggingface.co/avatars/802d923f053004ae894967693901882d.svg",
    "YOYO-AI": "https://huggingface.co/avatars/2c87edf1d12521c8bdb11c37bc9afee0.svg",
    "YXXPP": "https://huggingface.co/avatars/89379c9be09eb44f0cb8877f455ef966.svg",
    "Youssofal": "https://cdn-avatars.huggingface.co/v1/production/uploads/6957877a8cedfe83dba58b8d/N_00rxRoECw72WQ8IObdd.png",
    "abhishekchohan": "https://huggingface.co/avatars/9f11565724efe05e92937dc627a7869f.svg",
    "acon96": "https://huggingface.co/avatars/e784abe90576ef7ea4544938b4e9e6ff.svg",
    "aditya00196": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/2Jx7FET0z1EEQGkFwbr_O.png",
    "agemio": "https://huggingface.co/avatars/c3b2edf01560abe5117e1e5c8035223c.svg",
    "ai-server-mgr": "https://huggingface.co/avatars/0f37d60c3d380a34e7e8eb9e0f519339.svg",
    "ai4bharat": "https://cdn-avatars.huggingface.co/v1/production/uploads/1645373988267-5fbfc7a6e366524fe8e97ccf.png",
    "amd": "https://cdn-avatars.huggingface.co/v1/production/uploads/65ff7616871b36bf84150cda/mBg2Z4Vcdt5bkjxbYJFej.png",
    "amitdutta121": "https://huggingface.co/avatars/9e7e21169783a23425b51c47e1c3c0ee.svg",
    "antirez": "https://huggingface.co/avatars/0e9b8e0910a050085a1355fad6682439.svg",
    "aoiandroid": "https://cdn-avatars.huggingface.co/v1/production/uploads/64af86b446a29208da026734/KfJK5ZShc_TH8dVNSsbPf.png",
    "apple": "https://cdn-avatars.huggingface.co/v1/production/uploads/1653390727490-5dd96eb166059660ed1ee413.jpeg",
    "ashwin-sreedhar": "https://huggingface.co/avatars/940e62c4e6a4c94a8012deb9e2ee93ef.svg",
    "autospare": "https://cdn-avatars.huggingface.co/v1/production/uploads/69bdcba6cb655fc4ca714dfd/EJQ6jzIA3K2tmOJc6sc2W.jpeg",
    "aws": "https://cdn-avatars.huggingface.co/v1/production/uploads/66f19ed428ae41c20c470792/wtuzZxWzijQ3do3zYoOFH.png",
    "axolotl-ai-co": "https://cdn-avatars.huggingface.co/v1/production/uploads/641dfddf3bae5a77636817c5/Wa6Qn38GOAlhl6ClMv_Q3.png",
    "baa-ai": "https://cdn-avatars.huggingface.co/v1/production/uploads/644d2651d185572dd1e497d9/-F6ZUGFzgXLlFa1asCLBW.png",
    "baidu": "https://cdn-avatars.huggingface.co/v1/production/uploads/64f187a2cc1c03340ac30498/TYYUxK8xD1AxExFMWqbZD.png",
    "baseten": "https://cdn-avatars.huggingface.co/v1/production/uploads/66ba5a29e5904d821f2222f9/8QojWrmAHK1GMbcNOUvYn.png",
    "bb1975": "https://huggingface.co/avatars/74e2305ed5ed1fb0c53deba30f6e283a.svg",
    "beaupi": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/ZRTctcBYI8tYeibXOpSuW.png",
    "bharatgenai": "https://cdn-avatars.huggingface.co/v1/production/uploads/67b462a1f4f414c2b3e2bc2f/EnVeNWEIeZ6yF6ueZ7E3Y.jpeg",
    "black-forest-labs": "https://cdn-avatars.huggingface.co/v1/production/uploads/633f7a8f4be90e06da248e0f/m5YoF33abJ09vcwFxt1Mj.png",
    "bofenghuang": "https://huggingface.co/avatars/2a78c729e49a5669c49809f0e6dc0fbf.svg",
    "brainworkup": "https://cdn-avatars.huggingface.co/v1/production/uploads/660394aa593aa5f36af6367a/tFujO8S66A4bzy3W29yY_.webp",
    "build-small-hackathon": "https://cdn-avatars.huggingface.co/v1/production/uploads/60d2dc1007da9c17c72708f8/XejxNYtiveJxFYcILIEBh.png",
    "cPilotGod": "https://cdn-avatars.huggingface.co/v1/production/uploads/6a043c7f2d57400062e83bd6/hTAPe8TQkT4LR_JNHk_pL.png",
    "ccharnkij": "https://cdn-avatars.huggingface.co/v1/production/uploads/67e61df301dfc41992b97a54/qgEHhuM4IU0ZpY9uO7ohg.jpeg",
    "cerebras": "https://cdn-avatars.huggingface.co/v1/production/uploads/1678752782877-640fbb91bd54b1efde3678b2.png",
    "chabasx": "https://huggingface.co/avatars/a428d7af95dc72a3a0e13e8e770d7d06.svg",
    "chuanli11": "https://cdn-avatars.huggingface.co/v1/production/uploads/6223d06e2d6c7fd64778b805/0BgOvZTv-l4UMwjsBxkbE.jpeg",
    "cloudbjorn": "https://huggingface.co/avatars/3c46602429746b1bebbbb7e57f8d64ee.svg",
    "cloudyu": "https://cdn-avatars.huggingface.co/v1/production/uploads/64618bc5cf638aa8f856137c/zbUqrIeHjz41P3O2b3eey.jpeg",
    "colaiby2": "https://huggingface.co/avatars/5e639aabd0f44ebcaa1db78f2ad22afb.svg",
    "confamnode": "https://cdn-avatars.huggingface.co/v1/production/uploads/6929af9cf057f498adfd425a/q5YUSKDd4oWva_i9oAW40.png",
    "coqui": "https://cdn-avatars.huggingface.co/v1/production/uploads/1648459271425-61dbf25b99eac69222659467.png",
    "cs2764": "https://huggingface.co/avatars/2bc0d63b2e75ed3626ffff4c8171e5b0.svg",
    "cstr": "https://huggingface.co/avatars/5bab2253335e305c773730a0ac4bf66f.svg",
    "darkc0de": "https://cdn-avatars.huggingface.co/v1/production/uploads/6540a02d1389943fef4d2640/v9bXR0zJo6Cp7_H-RVPXh.png",
    "dealignai": "https://cdn-avatars.huggingface.co/v1/production/uploads/699af30d096f7e8fe8d82a11/PEWX90_WdOgBjRvqhsHix.png",
    "deejayprime": "https://cdn-avatars.huggingface.co/v1/production/uploads/6891cca78e06f503da0d07a0/SIrrDJCYtnPYhV56MD1Nj.png",
    "deepseek-ai": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
    "deepseek-community": "https://cdn-avatars.huggingface.co/v1/production/uploads/637ced166231c36c63707cba/ID37chRhBZ5vyljtMVTGn.png",
    "depth-anything": "https://www.gravatar.com/avatar/f4f72407f9653fd54129eee82ed335fa?d=retro&size=100",
    "dervig": "https://huggingface.co/avatars/31e1fb574adbc514b70021518d2c3366.svg",
    "dhairyashil": "https://huggingface.co/avatars/af8c802ab0c653238415cf0723adcfef.svg",
    "divinetribe": "https://cdn-avatars.huggingface.co/v1/production/uploads/6901e75af65f47018ef58191/x_a4n8RyEgj8fgO9x5ZCd.png",
    "dshive": "https://huggingface.co/avatars/c85d8b57f2cf353381d15c245ee4f449.svg",
    "edwixx": "https://cdn-avatars.huggingface.co/v1/production/uploads/65340cb69860c1cb37c8beb1/eHU9g-wEwX3P_2CkTbyRS.png",
    "erenyeager-1": "https://cdn-avatars.huggingface.co/v1/production/uploads/689da0b5751177485e1d874c/5K1TvedXBTlD8Z2XD7uMt.jpeg",
    "facebook": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
    "fantaxy": "https://cdn-avatars.huggingface.co/v1/production/uploads/66333c7887ce9a8935ff5738/j8x7UWos6XGf6SYuxD7B6.jpeg",
    "freestuff2875": "https://huggingface.co/avatars/78e2ec0933bf4712a2055a1001b32f33.svg",
    "fribourgquants": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/-cZrnO99BKUFhdNOV8vs6.png",
    "ggml-org": "https://cdn-avatars.huggingface.co/v1/production/uploads/63148d3b996c52bf0142cdbe/HXyNkyB0_nHI5WDNdiKHZ.png",
    "goodmai": "https://huggingface.co/avatars/595933a92c89d366d4bd90f3823dc86a.svg",
    "google": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
    "h2oai": "https://cdn-avatars.huggingface.co/v1/production/uploads/1673570796144-63c0a91f7f52541dfc7c831b.png",
    "hakunokina": "https://huggingface.co/avatars/732bb168da35842abc9cc95d319e0726.svg",
    "heni86": "https://cdn-avatars.huggingface.co/v1/production/uploads/6327f49875512645dec95c8b/tQJXNxFZ0icIeMwpwKxm_.png",
    "heretic-org": "https://cdn-avatars.huggingface.co/v1/production/uploads/68ce0ae4ce1399b97e364af4/vKbYrhGGUkC2O4YF0cF4_.png",
    "huihui-ai": "https://cdn-avatars.huggingface.co/v1/production/uploads/66228dc4c8920ec3513dc81a/9stczijVqp6TgQduzT8kN.png",
    "hunyuanvideo-community": "https://www.gravatar.com/avatar/c98615f99e27d862af2babbd9fbaa72c?d=retro&size=100",
    "hutangzhu": "https://huggingface.co/avatars/9b0c165913b6657ca9343dc0f0c0b8e6.svg",
    "hyperspaceai": "https://cdn-avatars.huggingface.co/v1/production/uploads/65cce14df0cac465b1f9f27b/nmACpqa5UL88sIOEtuv9l.png",
    "ibm-granite": "https://cdn-avatars.huggingface.co/v1/production/uploads/639bcaa2445b133a4e942436/CEW-OjXkRkDNmTxSu8Egh.png",
    "ibm-research": "https://cdn-avatars.huggingface.co/v1/production/uploads/637bfdf60dc13843b468ac20/npxapKcW-cXX3J2JBl2vY.png",
    "inference-optimization": "https://cdn-avatars.huggingface.co/v1/production/uploads/64ee5432f9ed75901c09a590/odGpnOKZOmkCqch45sAbm.png",
    "intel-ai": "https://cdn-avatars.huggingface.co/v1/production/uploads/632c2ffcdbea00ca21385b66/SecFPIgkn5o2ehWzR6HoH.png",
    "jackjohn001": "https://huggingface.co/avatars/c051e1331bd08e98b3d9a9ba76d4da3c.svg",
    "jc2375": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/YGYqKSh0aHMXIb5po5fKq.png",
    "jduartedj": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/Aq0a-WmyJ2sCF7rC1SOn1.png",
    "jgalego": "https://cdn-avatars.huggingface.co/v1/production/uploads/61d6fc8e7a9d6cf2044c2a32/nlPYjL9zY9raUfm8WCIo7.jpeg",
    "jtl11": "https://huggingface.co/avatars/37e5bec276240e8ea3515f48e5c6a593.svg",
    "jukofyork": "https://cdn-avatars.huggingface.co/v1/production/uploads/65995c45539c808e84c38bf1/QYhoTdzssz6psBAXJRoTp.png",
    "kaggle": "https://cdn-avatars.huggingface.co/v1/production/uploads/1655980755423-5fbfd09ee366524fe8e97cd3.png",
    "kepom": "https://huggingface.co/avatars/6e84c9c3cb48f5294c5c8a29363eee4e.svg",
    "khazarai": "https://cdn-avatars.huggingface.co/v1/production/uploads/675c41521363c2ddf38a9bd0/vlhrfwmUvhNKggXhfyFRg.jpeg",
    "kimi000": "https://huggingface.co/avatars/a6c4086fab8885af1d7ddd0a04703549.svg",
    "kmouratidis": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/VaviD0heHYH4y2ZyGOopk.png",
    "kostakoff": "https://cdn-avatars.huggingface.co/v1/production/uploads/684cc105c94f57af8b956042/VVOptoGetGoHuKAckrkEx.png",
    "koute": "https://huggingface.co/avatars/6742ac40002f7cd538382fc457a92a79.svg",
    "kp-forks": "https://www.gravatar.com/avatar/ec3ccf9bf953124b748e1ae3e03fde5d?d=retro&size=100",
    "krutrim-ai-labs": "https://cdn-avatars.huggingface.co/v1/production/uploads/679e056019b9b76439a55a33/xJACZDT7g3veiRGBvfXOK.png",
    "kszlachcikowski": "https://huggingface.co/avatars/f98bafeffd3ee9a3ab4ec40785ea3d61.svg",
    "kthierbach": "https://huggingface.co/avatars/ea34128be03bed63d2e01cf1b789cc03.svg",
    "kwanhee": "https://huggingface.co/avatars/fbdb6e9cc5e5e618db85bbc255c24eee.svg",
    "kyutai": "https://cdn-avatars.huggingface.co/v1/production/uploads/6355a3c1805be5a8f30fea49/8xGdIOlfkopZfhbMitw_k.jpeg",
    "lemuralabs": "https://cdn-avatars.huggingface.co/v1/production/uploads/6a43eaa1e610ba5400625802/fct9hOhOa2VVlgBjj4uA8.png",
    "lightseekorg": "https://cdn-avatars.huggingface.co/v1/production/uploads/6922906cab3f35a0141d405c/kJX1-rGiKoYtHWtHknnLA.png",
    "lite-infer": "https://cdn-avatars.huggingface.co/v1/production/uploads/655647bf2f76548766fce60f/sDrZMAm2YIZRg5_TdpNcG.png",
    "littlecosmos1990": "https://huggingface.co/avatars/402139590ab89a115d2051f20040d2d0.svg",
    "lkevincc0": "https://cdn-avatars.huggingface.co/v1/production/uploads/6831783766bc001411123e5e/zv5KAc34wUTkA356QM8VX.png",
    "llmfan46": "https://cdn-avatars.huggingface.co/v1/production/uploads/68851b893b66feaa5ca027d5/pxwPI83JFkYGWh5zD2r_c.png",
    "lm-kit": "https://cdn-avatars.huggingface.co/v1/production/uploads/648d6637e7f5198a203fd8f1/p7incVtEk7susYBjxTCuD.png",
    "lmstudio-community": "https://cdn-avatars.huggingface.co/v1/production/uploads/64445e5f1bc692d87b27e183/Quwh3tlKXa1y7aduVfxZi.png",
    "lmsys": "https://cdn-avatars.huggingface.co/v1/production/uploads/62d35f3ceaf3858ce253ab7a/uZ2jFNAjXfdBo6c6Yojmg.jpeg",
    "lordx64": "https://cdn-avatars.huggingface.co/v1/production/uploads/65510e83a12178dd75d63d5e/v3RXY9GzaU8w-NBfm7dS_.jpeg",
    "lvyufeng": "https://cdn-avatars.huggingface.co/v1/production/uploads/1647268601681-621a3baa2252352ffeb63af1.jpeg",
    "magespace": "https://cdn-avatars.huggingface.co/v1/production/uploads/62fbe33bc776fd8821ae77d8/WQkFHiIhizZxTpleyYHDC.png",
    "mattbucci": "https://huggingface.co/avatars/0b405b85f51402285629b3e80fafa878.svg",
    "mayflowergmbh": "https://cdn-avatars.huggingface.co/v1/production/uploads/609b8f1e7c3ca93d645cfca7/nMEyKsyWVD34nA-w4zR84.png",
    "medismera": "https://huggingface.co/avatars/eeae4a13634442725136b8abe97b96d2.svg",
    "meta-llama": "https://cdn-avatars.huggingface.co/v1/production/uploads/646cf8084eefb026fb8fd8bc/oCTqufkdTkjyGodsx1vo1.png",
    "meta-models": "https://cdn-avatars.huggingface.co/v1/production/uploads/64c115d1f661ace7c4022432/EbzfQisFJBEcKWDWLyV5O.png",
    "michaelfeil": "https://cdn-avatars.huggingface.co/v1/production/uploads/644fac0ce1d7a97f3b653ab1/qOpqx_bcy7RKkIrExtKrN.jpeg",
    "michaellu88": "https://huggingface.co/avatars/e87240d7d1c7c62bacaa9a12211c0683.svg",
    "microperceptron": "https://cdn-avatars.huggingface.co/v1/production/uploads/67eb650afb5d7fa30ec01229/FwzPZ6Ty2OO4c0Bn7LFKp.jpeg",
    "microsoft": "https://cdn-avatars.huggingface.co/v1/production/uploads/1583646260758-5e64858c87403103f9f1055d.png",
    "mingxilei": "https://huggingface.co/avatars/f750d88079fbb685f3229b20bac341bd.svg",
    "mirfan899": "https://cdn-avatars.huggingface.co/v1/production/uploads/1675348669572-606ffe7469a66931a0273ef5.jpeg",
    "mistralai": "https://cdn-avatars.huggingface.co/v1/production/uploads/634c17653d11eaedd88b314d/9OgyfKstSZtbmsmuG8MbU.png",
    "mlx-community": "https://cdn-avatars.huggingface.co/v1/production/uploads/623c830997ddced06d78699b/3qTjC7d3YFCJTwpxd2noq.png",
    "models123": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/OMZT6zYZu0BV1dPgAcK8Q.png",
    "mohammad2928git": "https://huggingface.co/avatars/c228585bc5eff35f1554daf520e83bfe.svg",
    "mondk": "https://cdn-avatars.huggingface.co/v1/production/uploads/6a0bf887438648586b82f731/6LirNjo5Y7d9swDCjWE9T.png",
    "moonshotai": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
    "mradermacher": "https://huggingface.co/avatars/96d0c12e8eb411b44ecefc1b6508a420.svg",
    "mrdbourke": "https://cdn-avatars.huggingface.co/v1/production/uploads/1659922312540-610a70f35a40a8bfebfbf09b.jpeg",
    "mrm8488": "https://cdn-avatars.huggingface.co/v1/production/uploads/5e4318d616b09a31220980d6/24rMJ_vPh3gW9ZEmj64xr.png",
    "myshell-ai": "https://cdn-avatars.huggingface.co/v1/production/uploads/641de0213239b631552713e4/LF00eHlddZADv5RRw1gJA.png",
    "nanonets": "https://cdn-avatars.huggingface.co/v1/production/uploads/641fc216a390e539522d511f/Xtxh40e8zSzkuKtCr58DH.jpeg",
    "nbkcnil": "https://huggingface.co/avatars/f9387d0de88f54a189c061c52e0166ca.svg",
    "nerkyor": "https://cdn-avatars.huggingface.co/v1/production/uploads/69ec923c8bdc19557f58c3e8/CRJ4-EeeA0J9pB_zWqKOi.png",
    "newtechdevng": "https://huggingface.co/avatars/cafd787f11b89a1d9ca2d10cd10c175a.svg",
    "nex-agi": "https://cdn-avatars.huggingface.co/v1/production/uploads/65435cad429b80b14922ab8d/a_O9jT_daz_NXTfxtcw6S.png",
    "ngabr": "https://huggingface.co/avatars/74677f5e8ff0d096a296545797a51a3d.svg",
    "ngoan": "https://huggingface.co/avatars/c4f60bdaa0edef2064b91783652341a4.svg",
    "nikita-savelyev-cerebras": "https://cdn-avatars.huggingface.co/v1/production/uploads/6963d5822c48c5aee514e0de/xsfsNmpl4-Z9qFdRHnRQa.jpeg",
    "nm-testing": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/qQp5Mx0lVZQLcA4ZKPi40.png",
    "np-deploys": "https://huggingface.co/avatars/b56c7198a1815cf2a9be3b4521843a8b.svg",
    "npario": "https://huggingface.co/avatars/2c34039153e198eef02663884b49acfd.svg",
    "nvidia": "https://cdn-avatars.huggingface.co/v1/production/uploads/65df9200dc3292a8983e5017/Vs5FPVCH-VZBipV3qKTuy.png",
    "onnx-community": "https://cdn-avatars.huggingface.co/v1/production/uploads/61b253b7ac5ecaae3d1efe0c/BpiYJ-m5mXPXfjkzEt6yZ.png",
    "open-unlearning": "https://cdn-avatars.huggingface.co/v1/production/uploads/6784a5ba30a231d476f50591/2NaB3ZkGc3NZ0yNoROesZ.png",
    "openai": "https://cdn-avatars.huggingface.co/v1/production/uploads/68783facef79a05727260de3/UPX5RQxiPGA-ZbBmArIKq.png",
    "openbmb": "https://cdn-avatars.huggingface.co/v1/production/uploads/1670387859384-633fe7784b362488336bbfad.png",
    "optimum-intel-internal-testing": "https://www.gravatar.com/avatar/9d73cfd35a0ce160f557f4b4e1d6d6a1?d=retro&size=100",
    "orcarouter": "https://cdn-avatars.huggingface.co/v1/production/uploads/65224256377bffdc599901c9/-qy82WgzPEAjof3QPGftY.png",
    "ovedrive": "https://huggingface.co/avatars/6a736e2e547b4782da96288a69d6cad2.svg",
    "p-e-w": "https://cdn-avatars.huggingface.co/v1/production/uploads/68ce0ae4ce1399b97e364af4/0sUhuSJaFvxPNSR8APc3V.png",
    "patrickbdevaney": "https://huggingface.co/avatars/f24ba3a80e566218af6dba046b5b4b90.svg",
    "peasantsmith": "https://cdn-avatars.huggingface.co/v1/production/uploads/6a477b974e15260d75a24a93/MZMzTCTprfdY24OxnMCL0.jpeg",
    "pr1daclar": "https://huggingface.co/avatars/e715371796601c36e62af9a5b95943fa.svg",
    "prince-canuma": "https://cdn-avatars.huggingface.co/v1/production/uploads/5fa241b4a13e063b8b2b5e2f/lbrO-eAcRDHqeoTPdMjkR.png",
    "prithivMLmods": "https://cdn-avatars.huggingface.co/v1/production/uploads/65bb837dbfb878f46c77de4c/23gZ_lBEwyoqjexFy9QLD.jpeg",
    "puwaer": "https://cdn-avatars.huggingface.co/v1/production/uploads/66e1be1d6d5a7f912ddb127a/cMYaawhwYTeHSeq275pIV.jpeg",
    "pyannote": "https://cdn-avatars.huggingface.co/v1/production/uploads/5fcf602140b27fad857dfa8d/-afjRExJpYCCl5TCtFA68.png",
    "qqceqqq": "https://huggingface.co/avatars/ad1d9d18139c8f1f0145d8246cab2041.svg",
    "rAVEUK": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/gyEC3OSXdYTyStDQ2wCAY.png",
    "reinforce20001": "https://huggingface.co/avatars/4d5ae8908d6d73d49fb35b373900ad66.svg",
    "ressl": "https://cdn-avatars.huggingface.co/v1/production/uploads/6a2de69a000819df3123924d/Bp7_mkenG63IOzmpW2NbF.png",
    "richardyoung": "https://cdn-avatars.huggingface.co/v1/production/uploads/6315fb2f29411a6864b05cea/ts85wrjXDDFBNdGJ1dXm0.jpeg",
    "rootlocalghost": "https://cdn-avatars.huggingface.co/v1/production/uploads/69c09fbb71675934ed60636f/uK3PbP0QRjbqxma-iv8aL.jpeg",
    "rtgfreuui": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/rkObz8zVZuSC7UGuxYsXg.png",
    "sahilchachra": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/pGAGIg1t8uJzK9Df4TG1W.png",
    "saidutta69": "https://cdn-avatars.huggingface.co/v1/production/uploads/636925bb1e9d04886d54fe30/2kSPWSfBV5L3IxJa3WJba.png",
    "sanjeevkumar70550932": "https://huggingface.co/avatars/0d40963bc43ad791530c2b5fdaf53696.svg",
    "saricles": "https://cdn-avatars.huggingface.co/v1/production/uploads/686c471fd23b104f7ea024c1/PZZ1TEyp99wciN2tBM2HM.jpeg",
    "sarvamai": "https://cdn-avatars.huggingface.co/v1/production/uploads/60270a7c32856987162c641a/umd13GCWVijwTDGZzw3q-.png",
    "scaryrawr": "https://huggingface.co/avatars/1390e61d60ef2fef5410478c55bd21a3.svg",
    "seanbailey518": "https://huggingface.co/avatars/6a0d92e15be5efd68fe715a8339de7b0.svg",
    "seawolf2357": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/H5ncB4vaBtP8GVCidgxL0.png",
    "segmind": "https://cdn-avatars.huggingface.co/v1/production/uploads/6335d72738e33ddc26527332/dMCE82gVz832ew-un3pqa.png",
    "sh0ck0r": "https://cdn-avatars.huggingface.co/v1/production/uploads/66e9018c3d096a4c8a957211/VQF53S2_Sq2p0EiyF4OL3.png",
    "shieldstackllc": "https://cdn-avatars.huggingface.co/v1/production/uploads/693351f356d0de9e2edd3c45/eqyDIbo65_v1gR5cHYxq-.png",
    "sigjhl": "https://huggingface.co/avatars/948953deb5e0295a73c7202e21df1f84.svg",
    "singersalt": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/SzLolQrJX9fUIdY4Y9eAX.png",
    "sjoerdbodbijl": "https://huggingface.co/avatars/db242e98796fd961c2a5caa53eaa66ac.svg",
    "skykraftstudios": "https://huggingface.co/avatars/f83a4d4a09c9a1cdcbe297372693db9b.svg",
    "sparkarena": "https://cdn-avatars.huggingface.co/v1/production/uploads/68efd01ad7f3daa413a3c0f5/vtyUcSAld4uKq_4YkzJUB.jpeg",
    "specsGuy": "https://huggingface.co/avatars/54ce51a37f8890f424075325295d739e.svg",
    "speechbrain": "https://cdn-avatars.huggingface.co/v1/production/uploads/1663000279893-60243f18c1f3c79f98e4b382.png",
    "spicyneuron": "https://huggingface.co/avatars/1b90869be21e6376a955383e7427c5ac.svg",
    "srswti": "https://cdn-avatars.huggingface.co/v1/production/uploads/6458b122c16ecb4815dd3ad7/fCvKXG5RNqmsAGGJKorTD.png",
    "sszymczyk": "https://huggingface.co/avatars/5f7db9ac59c9b755f23d4e2d10c6d8bf.svg",
    "stabilityai": "https://cdn-avatars.huggingface.co/v1/production/uploads/643feeb67bc3fbde1385cc25/7vmYr2XwVcPtkLzac_jxQ.png",
    "starrysproutkids": "https://huggingface.co/avatars/f25444cbf43b3776a4c2c648a628b35b.svg",
    "stepfun-ai": "https://cdn-avatars.huggingface.co/v1/production/uploads/644f7e6233ac8f46fa0b9e26/CmF2ocXhkr2UtHXgmwq7-.png",
    "swdq": "https://cdn-avatars.huggingface.co/v1/production/uploads/6468c937f43574d9556d0cd0/U0d0et6ucvTV4LlSi8WGt.jpeg",
    "tebruno99": "https://huggingface.co/avatars/62f594aeaf10fc3cd12dc652d66fc415.svg",
    "tencent": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/Lp3m-XLpjQGwBItlvn69q.png",
    "themindstudio": "https://cdn-avatars.huggingface.co/v1/production/uploads/67135e58d5df1a4dc6e05c44/pBAwdW4NcYadV0LxXDVl7.png",
    "therickcain": "https://huggingface.co/avatars/e0023813ccd62fa2b091fae205716e17.svg",
    "timteh673": "https://huggingface.co/avatars/c57b0dcb638b3dc8200cb028b700bcde.svg",
    "togethercomputer": "https://cdn-avatars.huggingface.co/v1/production/uploads/67be52be391885e452f4f1be/XiCaXuCIs-N_03X_nzopf.jpeg",
    "tonyloco2026": "https://huggingface.co/avatars/a910066ef33d36683f8060adda6c5184.svg",
    "tphage": "https://cdn-avatars.huggingface.co/v1/production/uploads/67bf05ab54eccd579b62ef23/3HUpMnMz-uJm2zx6Q-bMo.jpeg",
    "treadon": "https://cdn-avatars.huggingface.co/v1/production/uploads/6329c1a05c15412898abbf0c/HbLLtLhMOZXZonRvYVRFb.jpeg",
    "trohrbaugh": "https://cdn-avatars.huggingface.co/v1/production/uploads/no-auth/7b0Q-IgCRyocqTKxAWE9r.png",
    "trychoosing": "https://huggingface.co/avatars/5865772e6d927e3b03a55c8e290d150d.svg",
    "tvall43": "https://huggingface.co/avatars/5f2688bc2fdf77bb14fc3617426eb856.svg",
    "ubicloud": "https://huggingface.co/avatars/bf69f1e314ad49e7c3912ca7f7005c3d.svg",
    "ukisai": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/n9eOYEemFKMskG96VzaC0.png",
    "unaizulaika": "https://cdn-avatars.huggingface.co/v1/production/uploads/6512e5502a41af0d03d12fc9/vmqwgIlzvsvpR9UPDxrg6.jpeg",
    "unsloth": "https://cdn-avatars.huggingface.co/v1/production/uploads/62ecdc18b72a69615d6bd857/E4lkPz1TZNLzIFr_dR273.png",
    "ununununununununununununununununununun": "https://cdn-avatars.huggingface.co/v1/production/uploads/6460e9ec9aad9e149722e9ef/DqDb6pLUzI8GOo5R1PqL7.jpeg",
    "vastai-ais": "https://cdn-avatars.huggingface.co/v1/production/uploads/64b7afe4fa7eabaae503e244/O_tbH14R3tn62IaF-rGd0.png",
    "vcruz305": "https://cdn-avatars.huggingface.co/v1/production/uploads/674f5d48ccdcfbc40f75ac78/9_T645MbbUhzfZFPEPJPF.jpeg",
    "veeceey": "https://huggingface.co/avatars/a69e690e7f009848849965a03a33bca4.svg",
    "vftfnn": "https://huggingface.co/avatars/30f34f7d098a2be7f1cbf6e14fdca127.svg",
    "virtualdmns": "https://cdn-avatars.huggingface.co/v1/production/uploads/6419ae4ce8a6183a8cabf2f2/KY-Viv5CEbv1tgvW2ArRB.jpeg",
    "vpyn": "https://huggingface.co/avatars/d87b1476f3f34ea3480fd70a87a611ad.svg",
    "wangzhang": "https://huggingface.co/avatars/1813a55ddadca71a7502740d992a702c.svg",
    "wealthcoders": "https://huggingface.co/avatars/f8b427c3f9a339cfa8dafa9ee58576ed.svg",
    "whrrr": "https://huggingface.co/avatars/d178e469d61f77b4262f433714048fcf.svg",
    "win10": "https://cdn-avatars.huggingface.co/v1/production/uploads/640720285e6d06cc2cf346e4/p-U1ON7ZPNntBAnOo8FCE.png",
    "xcreates": "https://cdn-avatars.huggingface.co/v1/production/uploads/6930102f5a301d8b708c299c/Gu8PC05oCuRaUJ3BuAKu8.png",
    "xtools-at": "https://cdn-avatars.huggingface.co/v1/production/uploads/6a27f989cbc5de730449235d/Xz6fliMHyU-NQHX_ot0cJ.jpeg",
    "yifanyu": "https://huggingface.co/avatars/0c4111a5624181169783a58b0a381f37.svg",
    "ysharma": "https://cdn-avatars.huggingface.co/v1/production/uploads/1624431552569-noauth.jpeg",
    "z-lab": "https://cdn-avatars.huggingface.co/v1/production/uploads/650dac79b959b0e1d41d7378/VCcvseVUp0ImXA3r-gPIi.png",
    "zai-org": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
    "zero0303": "https://huggingface.co/avatars/3eb3c7c2c254259c97e5ef95fc89848c.svg",
    "zerofata": "https://cdn-avatars.huggingface.co/v1/production/uploads/65b19c6c638328850e12d38c/jQtnPYhgJFF62V-ch1wil.png",
    "zhoude": "https://huggingface.co/avatars/037d5c38c3e6837ac3c79538787bcf05.svg",
    "zhuyksir": "https://cdn-avatars.huggingface.co/v1/production/uploads/6630821b6ee021123fe91604/_4a4W7crTHIYncs04wPmw.jpeg",
    "zml5418": "https://huggingface.co/avatars/632b20ea8b2c1b3ec4ba3fa769bade7a.svg"
  },
  "price_list": {
    "usd_per_cpu_hour": 0.036,
    "usd_per_gb_ram_hour": 0.003,
    "usd_per_gpu_hour": 0.36,
    "usd_per_kwh": 0.1,
    "source": "epn-daemon config.DefaultPriceFloors: what metering charges every gram"
  },
  "models": [
    {
      "model": "Gemma 4 E2B",
      "made_by": "Google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/gemma-4/e2b",
      "hub_url": "https://huggingface.co/google/gemma-4-E2B-it",
      "download_gb": 3.8,
      "params_b": 5.1,
      "active_params_b": 2,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu",
        "Amharic",
        "Assamese",
        "Spanish",
        "Nepali",
        "Dutch",
        "Odia",
        "Russian",
        "Sinhala",
        "Swahili",
        "Urdu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux PC with an NVIDIA GPU, 256 GB",
        "machine_detail": null,
        "machine_class": "linux/amd64/nvidia/256g",
        "machines": 1,
        "tokens_per_sec": 291.17,
        "context_tokens": 131072,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": 1.6475,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates",
        "measured_at": "2026-11-03T02:29:57.083Z",
        "signed": true,
        "build_page": null,
        "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
        "weeks": [
          "2026-W45"
        ]
      },
      "cheapest": {
        "machine": "Linux PC with an NVIDIA GPU, 32 GB",
        "machine_detail": null,
        "machine_class": "linux/amd64/nvidia/32g",
        "machines": 2,
        "tokens_per_sec": 77.58,
        "context_tokens": 131072,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": 0.433,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates",
        "measured_at": "2026-11-02T21:55:45.037Z",
        "signed": true,
        "build_page": null,
        "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
        "weeks": [
          "2026-W43",
          "2026-W44",
          "2026-W45"
        ]
      },
      "machine_kinds": 24,
      "machines_total": 47,
      "weeks_measured": [
        "2026-W29",
        "2026-W31",
        "2026-W34",
        "2026-W35",
        "2026-W36",
        "2026-W37",
        "2026-W38",
        "2026-W40",
        "2026-W41",
        "2026-W42",
        "2026-W43",
        "2026-W44",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux PC with an NVIDIA GPU, 256 GB",
          "machine_detail": null,
          "machine_class": "linux/amd64/nvidia/256g",
          "machines": 1,
          "tokens_per_sec": 291.17,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 1.6475,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates",
          "measured_at": "2026-11-03T02:29:57.083Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W45"
          ]
        },
        {
          "machine": "Linux · RTX PRO 6000 Blackwell Server Edition 96 GB · 48 CPUs · 177 GB RAM",
          "machine_detail": {
            "name": "Linux · RTX PRO 6000 Blackwell Server Edition 96 GB · 48 CPUs · 177 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 48,
            "cpu_model": "AMD EPYC 9B45",
            "ram_gb": 177,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "RTX PRO 6000 Blackwell Server Edition",
            "gpu_count": 1,
            "vram_gb": 96,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|48|177|nvidia|96|RTX PRO 6000 Blackwell Server Edition|1|Google Colab"
          },
          "machine_class": "linux|amd64|48|177|nvidia|96|RTX PRO 6000 Blackwell Server Edition|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 291.17,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.0053,
          "cost_breakdown": {
            "tokens_per_sec": 291.17,
            "usd_per_hour": 2.1019,
            "usd_per_million": 2.0053,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "47 of 48 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 470,
                "usd_per_million": 1.6142
              },
              {
                "resource": "Memory",
                "quantity": "3.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.175,
                "usd_per_million": 0.0109
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.3434
              },
              {
                "resource": "Energy",
                "quantity": "385 W modelled (47 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 10.694,
                "usd_per_million": 0.0367
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 9B45",
              "model_raw": "AMD EPYC 9B45",
              "clock_ghz": null,
              "cores_logical": 48,
              "cores_physical": 24,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 32,
              "single_core_hashes_per_sec": 1937.9,
              "multi_core_hashes_per_sec": 70559.4,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-03T02:23:41.449Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 176.9,
              "bandwidth_gb_s": 26.2,
              "latency_ns": 12,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-03T02:23:41.985Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA RTX PRO 6000 Blackwell Server Edition",
              "memory_gb": 95,
              "fp16_tflops": 372.6,
              "watts": 39.8,
              "tflops_per_watt": 9.353,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-03T02:33:49.172Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 15977,
              "write_mb_s": 36.1,
              "random_read_iops": 1825761,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-03T02:23:43.974Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T02:42:33.308Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220cf803b3f11052307765669b9d23c5c214b142184538b81d7b2779d8cd1e189d2",
            "signature": "v0KKPp0Se3QbAB8vgwMmja9knVDNuEvglNHE4SuN2zcR3uZsX7LvcPmKUBW5Gev5eCacSafd/gCCa3jQGtf/BA==",
            "signing_payload_b64": "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",
            "payload_sha256": "KGFGnLsCiMhio27cBxMKafXeCyrjOYV+mUk/A0nZ+WM="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "e189d2"
          ]
        },
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": 182.88,
          "context_tokens": 4096,
          "build": "IQ4_XS (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_XS",
          "engine": "ollama",
          "cost_usd_per_million": 1.4139,
          "cost_breakdown": {
            "tokens_per_sec": 182.88,
            "usd_per_hour": 0.9308,
            "usd_per_million": 1.4139,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.8202
              },
              {
                "resource": "Memory",
                "quantity": "2.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 2.315,
                "usd_per_million": 0.0127
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.5468
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0342
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "RsptUcs06S28m0uKeWTlZS9QCPWg2PQHX0AsXewXagFWGj7jdYaQVULZ2Pf5WzK+wbsPNxE0LVc/oErokeeuAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "p8d6ff7bVFI3ZqIZCAqBkR+SgYhM4f1wn/n7zRTgckI="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-iq4_xs-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 170.45,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by google",
          "ref": "gemma4:e2b",
          "engine": "ollama",
          "cost_usd_per_million": 1.5375,
          "cost_breakdown": {
            "tokens_per_sec": 170.45,
            "usd_per_hour": 0.9435,
            "usd_per_million": 1.5375,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.88
              },
              {
                "resource": "Memory",
                "quantity": "7.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 5.828,
                "usd_per_million": 0.0342
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.5867
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0367
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "0jaTRjp3ghIwgAyLI6t5t/fAAM2aQDoaoB1Ea1xTGuE0ut1V/SB1fw/cqn99aTmDwN5aC77NLkARiTs7CQVFDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "j33xtQnznek8EvD5kEKzrJSGKY4HtWocgMM15i3H5TM="
          },
          "build_page": "/inferences/gemma-4/e2b/google-q4_k_m-ollama",
          "hub_url": null,
          "weeks": [
            "2026-W31",
            "2026-W34",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 149.89,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by google",
          "ref": "gemma4:e2b",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.4779,
          "cost_breakdown": {
            "tokens_per_sec": 149.89,
            "usd_per_hour": 0.7975,
            "usd_per_million": 1.4779,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 0.7339
              },
              {
                "resource": "Memory",
                "quantity": "7.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 5.828,
                "usd_per_million": 0.0389
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.6672
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.038
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "geBl5+OCuwaouiLNj4hTEXccZoUTUhxy8pspAzA0cbYywlce34JGlhJmlnjnSjjQld3udO9AClLy6gn+cht/CA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTQ5Ljg4ODE0NjIyMDg4NTEsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0MDI4MTkyOTcwOTcyMjA1LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwNTI0MDUwZTkyOTZkMzEyY2QyODI4YzQyNjE2NWJmZjIzYmFlZWY3NjlhZGE5NzA4MzA5ODgzMmYwNmRhZTY3OSIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbnZpZGlhLzY0ZyIsIm1vZGVsIjoiZ2VtbWE0OmUyYiIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo0NSwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjo3MC41OTkzMTE5OTkwMDAwMSwidG90YWxfdG9rZW5zIjoxMDU4Mn19",
            "payload_sha256": "KaXNHyevfmFklL4IbToF97deO3IIeXBfknigo3L9lmk="
          },
          "build_page": "/inferences/gemma-4/e2b/google-q4_k_m-ollama",
          "hub_url": null,
          "weeks": [
            "2026-W31",
            "2026-W34",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux PC with an NVIDIA GPU, 64 GB",
          "machine_detail": null,
          "machine_class": "linux/amd64/nvidia/64g",
          "machines": 1,
          "tokens_per_sec": 98.85,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 1.1604,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates",
          "measured_at": "2026-11-06T05:45:12.775Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W45"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 98.85,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.2141,
          "cost_breakdown": {
            "tokens_per_sec": 98.85,
            "usd_per_hour": 0.7879,
            "usd_per_million": 2.2141,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 1.1128
              },
              {
                "resource": "Memory",
                "quantity": "3.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.175,
                "usd_per_million": 0.0321
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.0116
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0576
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "0ojOPVeq3RHALFzo3fjit9ENkN5A6Kg+7QvwzadHx2LbjjPQGmU7IRgYm36SobK4PYQapZua3fjyupR/4J+tAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "629GRJoEP+hRBaykSjsTqOifOWURQv1eot/WIwrB+qs="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows PC with an NVIDIA GPU, 16 GB",
          "machine_detail": null,
          "machine_class": "windows/amd64/nvidia/16g",
          "machines": 2,
          "tokens_per_sec": 97.95,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 2.6485,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates",
          "measured_at": "2026-10-12T14:30:23.535Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W41",
            "2026-W42"
          ]
        },
        {
          "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "Tesla T4",
            "gpu_count": 2,
            "vram_gb": 15,
            "host": "Kaggle",
            "vendor_org": "nvidia",
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
          "machines": 1,
          "tokens_per_sec": 91.08,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by google",
          "ref": "gemma4:e2b",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.5416,
          "cost_breakdown": {
            "tokens_per_sec": 91.08,
            "usd_per_hour": 0.5055,
            "usd_per_million": 1.5416,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.3294
              },
              {
                "resource": "Memory",
                "quantity": "7.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 5.828,
                "usd_per_million": 0.064
              },
              {
                "resource": "GPU",
                "quantity": "1 of 2 cards (the one this build fits on)",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.098
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0503
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "531znaVOfUJSwuU7xv+fghDMIf2XaH4Mp9+ec4R259syh5BX4nZaFlgJYpFmDWgG7uuKivZI2+nI6VCQkrSgAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "QDi7SmOTdtP4NFpzBT3MLT4gBdmcMKjhNytNz6BEYIU="
          },
          "build_page": "/inferences/gemma-4/e2b/google-q4_k_m-ollama",
          "hub_url": null,
          "weeks": [
            "2026-W31",
            "2026-W34",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 87.36,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama",
          "cost_usd_per_million": 2.9697,
          "cost_breakdown": {
            "tokens_per_sec": 87.36,
            "usd_per_hour": 0.9339,
            "usd_per_million": 2.9697,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 1.7171
              },
              {
                "resource": "Memory",
                "quantity": "3.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.175,
                "usd_per_million": 0.0363
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.1447
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0715
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "g7D+Q6k6zJhxTSi+PBsK06ZcHIejAEnTvClNj43djf0L+uTIb2uN4qKfakhrxurLDQRxh6BkvlT7HDWhndFNDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "7/KuC28eu6TwKMyb+IJy6dUVh1xPI/0rSPXR1DOwQF0="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 83.82,
          "context_tokens": 131072,
          "build": "Q4_0 (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.6269,
          "cost_breakdown": {
            "tokens_per_sec": 83.82,
            "usd_per_hour": 0.4909,
            "usd_per_million": 1.6269,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.3579
              },
              {
                "resource": "Memory",
                "quantity": "2.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.777,
                "usd_per_million": 0.0212
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.1931
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0547
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "AFpY9+ffmcniz6Nfj3L9Jv+dM5zaMV9OC+iSbg3xPqxhsmkcpXRBfSGoUzH7920KsQ39B1O2ufNqqWKvcG2QAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "Lv5VaZ6/OgBL4iAb68CoDGUhBhUYlQSN3A7eEKpmVlU="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W35",
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "Tesla T4",
            "gpu_count": 2,
            "vram_gb": 15,
            "host": "Kaggle",
            "vendor_org": "nvidia",
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
          "machines": 1,
          "tokens_per_sec": 78.69,
          "context_tokens": 131072,
          "build": "IQ4_XS (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_XS",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.7397,
          "cost_breakdown": {
            "tokens_per_sec": 78.69,
            "usd_per_hour": 0.4928,
            "usd_per_million": 1.7397,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.3812
              },
              {
                "resource": "Memory",
                "quantity": "2.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 2.315,
                "usd_per_million": 0.0294
              },
              {
                "resource": "GPU",
                "quantity": "1 of 2 cards (the one this build fits on)",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.2708
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0582
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "UXXrqrsUXrQW86qz+sZdY2gIaJUOd2bt+qdnN0tfvxeA+LrYrQDmM43UIx2V8HaFWnzPKflXmz3RixcOKp1rDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "+SJCD5FFktWC4iZk9/sbteddYsESplj6RpN9Pwdl5IQ="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-iq4_xs-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux PC with an NVIDIA GPU, 32 GB",
          "machine_detail": null,
          "machine_class": "linux/amd64/nvidia/32g",
          "machines": 2,
          "tokens_per_sec": 77.58,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 0.433,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates",
          "measured_at": "2026-11-02T21:55:45.037Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 2,
          "tokens_per_sec": 77.58,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.7757,
          "cost_breakdown": {
            "tokens_per_sec": 77.58,
            "usd_per_hour": 0.4959,
            "usd_per_million": 1.7757,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.3867
              },
              {
                "resource": "Memory",
                "quantity": "3.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.175,
                "usd_per_million": 0.0409
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.289
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0591
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-19T11:28:03.450Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:00240801122095368b040a04cc8b46563ce9ac25e091edccd520d7753bb0a0093f9490e78fd2",
            "signature": "OVe3nttZDdDnZkX344L+tBDQQsFslXcX4T6fYZDm8AXTtAE2Ki6wp2w1ikvYhW/kC46HbqrTXOjoz9dpmpVQCQ==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NzcuNTc3NTI3NDExNDA4MTgsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkyNDA5MjgzNDUwMTYxODkyLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwOTUzNjhiMDQwYTA0Y2M4YjQ2NTYzY2U5YWMyNWUwOTFlZGNjZDUyMGQ3NzUzYmIwYTAwOTNmOTQ5MGU3OGZkMiIsImV4dHJhIjp7Im1vZGVsIjoiaGYuY28vdW5zbG90aC9nZW1tYS00LUUyQi1pdC1HR1VGOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50IjozNSwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjo5Ny43NzMxNjAwMDAwMDAwMiwidG90YWxfdG9rZW5zIjo3NTg1fX0=",
            "payload_sha256": "Yn8g7FG6ZJeVRPyQI1hfyHm9MwQeGWZQjYqWcxW1z8M="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "e78fd2",
            "a5be46"
          ]
        },
        {
          "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "Tesla T4",
            "gpu_count": 2,
            "vram_gb": 15,
            "host": "Kaggle",
            "vendor_org": "nvidia",
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
          "machines": 1,
          "tokens_per_sec": 74.18,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.8572,
          "cost_breakdown": {
            "tokens_per_sec": 74.18,
            "usd_per_hour": 0.4959,
            "usd_per_million": 1.8572,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.4044
              },
              {
                "resource": "Memory",
                "quantity": "3.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.175,
                "usd_per_million": 0.0428
              },
              {
                "resource": "GPU",
                "quantity": "1 of 2 cards (the one this build fits on)",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.3482
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0618
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 312.8,
              "multi_core_hashes_per_sec": 618.9,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T21:33:01.078Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 2.4,
              "latency_ns": 21.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T21:33:01.673Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 21.6,
              "watts": 59.1,
              "tflops_per_watt": 0.365,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-02T21:43:16.805Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4293.1,
              "write_mb_s": 60.4,
              "random_read_iops": 514568,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-02T21:33:02.973Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T00:21:51.818Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220d79e72fae9f23f0437b58a3411e6bdb14e5c77f568cc3bdf6622534463c76d30",
            "signature": "+zNCrJNhpfdBojlMeA+Mp0j/4evVbY618zdYeDdYAGc85a9yuXO3x5EXwQGwZPUP5oF04a5qgpuMOxqPHp4wAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "dVGqqLD+hVl/M/Km3YocSTqlIWfZVsI88YNfyrlk8ic="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "c76d30"
          ]
        },
        {
          "machine": "Mac · no GPU · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|none|0|||"
          },
          "machine_class": "darwin|arm64|10|16|none|0|||",
          "machines": 1,
          "tokens_per_sec": 57.55,
          "context_tokens": 4096,
          "build": "Q4_0 (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.6164,
          "cost_breakdown": {
            "tokens_per_sec": 57.55,
            "usd_per_hour": 0.3349,
            "usd_per_million": 1.6164,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 1.5638
              },
              {
                "resource": "Memory",
                "quantity": "2.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.777,
                "usd_per_million": 0.0309
              },
              {
                "resource": "Energy",
                "quantity": "45 W modelled (9 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 1.25,
                "usd_per_million": 0.0217
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "X6xSFl9Do3o55VvjwVJO5dOIn2c1W1u3LYxcNXH3CdqPx6/4HXARa+zHBV+Y5qjPXjy0Kk1iwNGSQICwml9oCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "A3KKwoaT+bo1eMQR1EiO0CL8jaH1MA/+Pv6NR8OThUw="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W35",
            "2026-W43"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Mac with Apple silicon, 16 GB",
          "machine_detail": null,
          "machine_class": "darwin/arm64/apple/16g",
          "machines": 1,
          "tokens_per_sec": 53.69,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 1.381,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates",
          "measured_at": "2026-11-02T09:23:57.166Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W45"
          ]
        },
        {
          "machine": "Mac · Apple M4 · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple M4 · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": "Apple M4",
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": "Apple M4",
            "gpu_count": 1,
            "vram_gb": null,
            "host": "own machine",
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|Apple M4|1|own machine"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|Apple M4|1|own machine",
          "machines": 1,
          "tokens_per_sec": 53.69,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "cost_usd_per_million": 3.6987,
          "cost_breakdown": {
            "tokens_per_sec": 53.69,
            "usd_per_hour": 0.7149,
            "usd_per_million": 3.6987,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 1.6762
              },
              {
                "resource": "Memory",
                "quantity": "3.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.175,
                "usd_per_million": 0.0591
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.8625
              },
              {
                "resource": "Energy",
                "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.417,
                "usd_per_million": 0.1009
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "UAbEVKskbftiXCBPpq98EKE67ccZSxX28i8PpJuhscmz+xI/DaBYd5o20oOxPsx3sHE37vDjkB8yZ7/8obRRCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "GuAzHNGJNLDo46XCZWcy7SexOTRTk66Ra3CObYxU5s0="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Windows · NVIDIA 4 GB · 12 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 4 GB · 12 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 12,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 4,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|12|16|nvidia|4|||"
          },
          "machine_class": "windows|amd64|12|16|nvidia|4|||",
          "machines": 1,
          "tokens_per_sec": 28.05,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "cost_usd_per_million": 7.8028,
          "cost_breakdown": {
            "tokens_per_sec": 28.05,
            "usd_per_hour": 0.7879,
            "usd_per_million": 7.8028,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 3.9216
              },
              {
                "resource": "Memory",
                "quantity": "3.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.175,
                "usd_per_million": 0.1132
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 3.5651
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.203
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i7-11600H",
              "model_raw": "11th Gen Intel Core i7-11600H",
              "clock_ghz": null,
              "cores_logical": 12,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 365.3,
              "multi_core_hashes_per_sec": 3482.5,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-01T21:15:49.919Z",
              "measured_in": "IN_560037",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.2,
              "latency_ns": 44.2,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-01T21:15:50.572Z",
              "measured_in": "IN_560037",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3050 Laptop GPU",
              "memory_gb": 4,
              "fp16_tflops": 0,
              "watts": null,
              "tflops_per_watt": null,
              "method": null,
              "measured_at": "2026-09-11T19:24:20.219Z",
              "measured_in": null,
              "signed": true
            },
            "storage": {
              "read_mb_s": 1409.8,
              "write_mb_s": 340.5,
              "random_read_iops": 46320,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-01T21:15:51.083Z",
              "measured_in": "IN_560037",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-01T21:35:20.666Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201844bc3835f475fd8afa64d04ddcb9a1e0087bb40a31973d91b193e6c953b229",
            "signature": "Gpwv/X8hmzFtV/ZcQ2e7QmNchYVdBaKyv/X31LdiK/QVFcF7KUVK6fxUF5u6ZrXQbQjpb91fZtsGsTlBmIoFAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "6A5NLcqkeGMVjP7JR875olLeAK09QSJT1ezPwpRxHi4="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "53b229"
          ]
        },
        {
          "machine": "Linux PC with no GPU, 32 GB",
          "machine_detail": null,
          "machine_class": "linux/amd64/none/32g",
          "machines": 1,
          "tokens_per_sec": 24.89,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 1.3496,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates",
          "measured_at": "2026-10-27T00:42:02.167Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W44"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 24.89,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.3495,
          "cost_breakdown": {
            "tokens_per_sec": 24.89,
            "usd_per_hour": 0.1209,
            "usd_per_million": 1.3495,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.2052
              },
              {
                "resource": "Memory",
                "quantity": "3.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.175,
                "usd_per_million": 0.1275
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0168
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "xoMYQdk2S6coNrQoSe48y4BUZ1DYdZQPvF/hlFNFm69VxgaaL9EQswDkBgX8bbbmrI1J40aGYzNpQDY18pk8DA==",
            "signing_payload_b64": "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",
            "payload_sha256": "+XON2DT74QDrtvHe/FF4lw7Xl04ZAIijAFNy+D2QUhw="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Windows PC with an NVIDIA GPU, 8 GB",
          "machine_detail": null,
          "machine_class": "windows/amd64/nvidia/8g",
          "machines": 2,
          "tokens_per_sec": 15.33,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 9.6476,
          "cost_is_estimate": true,
          "cost_basis": "the network's metered rates, machine size assumed",
          "measured_at": "2026-09-16T11:18:57.277Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38"
          ]
        },
        {
          "machine": "Windows · NVIDIA 4 GB · 12 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 4 GB · 12 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 12,
            "cpu_model": null,
            "ram_gb": 7,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 4,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|12|7|nvidia|4|||"
          },
          "machine_class": "windows|amd64|12|7|nvidia|4|||",
          "machines": 1,
          "tokens_per_sec": 15.33,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama",
          "cost_usd_per_million": 14.2764,
          "cost_breakdown": {
            "tokens_per_sec": 15.33,
            "usd_per_hour": 0.7879,
            "usd_per_million": 14.2764,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 7.1751
              },
              {
                "resource": "Memory",
                "quantity": "3.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.175,
                "usd_per_million": 0.2071
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 6.5228
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.3714
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-22T10:54:02.459Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220b3960d56a82de98ec563fb48bbfc3ca6b4dc788b19545f4315f8f363aa9a1db0",
            "signature": "BKh6EJDNuj5VCyiwsFSToaK8T+f+QRBFAB4GsKdgvONp8CJL68IfgeJ7ADUT0cC8vnBgYAA66/mtlTjz5D3PDg==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTUuMzMwODcwNDc3MjQ2NzM5LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5MDA3NDQ0MjQ1OTUyNDIwMCwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMGIzOTYwZDU2YTgyZGU5OGVjNTYzZmI0OGJiZmMzY2E2YjRkYzc4OGIxOTU0NWY0MzE1ZjhmMzYzYWE5YTFkYjAiLCJleHRyYSI6eyJtb2RlbCI6ImhmLmNvL3Vuc2xvdGgvZ2VtbWEtNC1FMkItaXQtR0dVRjpRNF9LX00iLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEiLCJzYW1wbGVfY291bnQiOjEwLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjExNi4zNjY1MTY5OTkwMDAwMSwidG90YWxfdG9rZW5zIjoxNzg0fX0=",
            "payload_sha256": "8mu9Eb6jJxaeXAcubqD5SOFwPqaAdreeRWaSMkJ1Y80="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "9a1db0"
          ]
        },
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": 13,
          "context_tokens": 4096,
          "build": "Q3_K_S (3-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q3_K_S",
          "engine": "ollama",
          "cost_usd_per_million": 5.5982,
          "cost_breakdown": {
            "tokens_per_sec": 13,
            "usd_per_hour": 0.2619,
            "usd_per_million": 5.5982,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 5.3866
              },
              {
                "resource": "Memory",
                "quantity": "2.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.777,
                "usd_per_million": 0.1367
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0748
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "ciAHBLyGB79ij368pSfUFAczmAnHBePLvRuSY/iXnGm7RZFvDexxcDWgQ1DMTP84U2OLlgoXQK6suaIEWN/eCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "HEMTO6qC/psFOEv69Hs+ibyhavMscQB5YyFsCNa1bbw="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W36",
            "2026-W37"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Windows · NVIDIA 4 GB · 8 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 4 GB · 8 CPUs · 8 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 4,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|8|8|nvidia|4|||"
          },
          "machine_class": "windows|amd64|8|8|nvidia|4|||",
          "machines": 1,
          "tokens_per_sec": 9.95,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama",
          "cost_usd_per_million": 17.9178,
          "cost_breakdown": {
            "tokens_per_sec": 9.95,
            "usd_per_hour": 0.6419,
            "usd_per_million": 17.9178,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 7.0339
              },
              {
                "resource": "Memory",
                "quantity": "3.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.175,
                "usd_per_million": 0.319
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 10.0485
              },
              {
                "resource": "Energy",
                "quantity": "185 W modelled (7 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.139,
                "usd_per_million": 0.5164
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-26T10:45:15.368Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220df04c29e6a0ad2ef0dc554f22fdbb03346428fde27cc62255d9dd9c61412954d",
            "signature": "7ghbLIM0mYzc7+MMSlVzUcXO9oRynEgJKiNgmYUASmPyB7MLJgr762sG79s2MDqKFWMAQCKPN893YejivYy9Bw==",
            "signing_payload_b64": "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",
            "payload_sha256": "lVrR+u4k4x1+bGFTjgCnr1O6OldbXHw8jAyvI7fgT50="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "12954d"
          ]
        },
        {
          "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "windows|amd64|8|16|none|0|||"
          },
          "machine_class": "windows|amd64|8|16|none|0|||",
          "machines": 1,
          "tokens_per_sec": 9.1,
          "context_tokens": 4096,
          "build": "Q4_0 (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_0",
          "engine": "ollama",
          "cost_usd_per_million": 7.9949,
          "cost_breakdown": {
            "tokens_per_sec": 9.1,
            "usd_per_hour": 0.2619,
            "usd_per_million": 7.9949,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 7.6928
              },
              {
                "resource": "Memory",
                "quantity": "2.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.777,
                "usd_per_million": 0.1953
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.1068
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "C1x1pWyS4Hl82iIs75U95aAu5WSkwkVDcdmJr7T0FfNrm8lXjp/7vjh4IyIH1n0xpFcDdvuiqVt18a2N0p+vBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "yZ/NXHXpTosnejliBa+6+KvbkNJv7jc7lMPCzd2Qpyo="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W35",
            "2026-W43"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 7.15,
          "context_tokens": 8192,
          "build": "Q4_K_M (4-bit) by google",
          "ref": "gemma4:e2b",
          "engine": "ollama-host",
          "cost_usd_per_million": 5.0695,
          "cost_breakdown": {
            "tokens_per_sec": 7.15,
            "usd_per_hour": 0.1305,
            "usd_per_million": 5.0695,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 4.196
              },
              {
                "resource": "Memory",
                "quantity": "7.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 5.828,
                "usd_per_million": 0.8151
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0583
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "XryWV44xDp4vYZ6l4mKLHXmvcI147mlFr5Ff5/t7TuiAcIrL3JS4PsRIp+1/x15DXGzuHQehrtc5rMiEb1vFCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "VzaYwMVgqCqeX66cmd21g+qey9AXFL3q04dTRasJH0s="
          },
          "build_page": "/inferences/gemma-4/e2b/google-q4_k_m-ollama",
          "hub_url": null,
          "weeks": [
            "2026-W31",
            "2026-W34",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 6.93,
          "context_tokens": 4096,
          "build": "IQ4_XS (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_XS",
          "engine": "ollama-host",
          "cost_usd_per_million": 4.7232,
          "cost_breakdown": {
            "tokens_per_sec": 6.93,
            "usd_per_hour": 0.1178,
            "usd_per_million": 4.7232,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 4.329
              },
              {
                "resource": "Memory",
                "quantity": "2.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 2.315,
                "usd_per_million": 0.3341
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0602
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "APPN3tO3Hy3RRDM5bpPvLeAadDpcGc5M/dGvzK0TtEb3Ji0Z/MRDwdsbfRWRgXkLn+00w4FThJc2tG4GfwJFBQ==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6Ni45MzAwNTA3OTk5ODgwOTYsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzNDA1OTE3Mjk4NjQ4ODkxLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwZWUzNWRjNDZhMjIxMmNiMzE0NDRjYTRjMTk2NjlhYzhkOTdiMDkyNTIwYjhjOWU0ZDI5YWI0OGQ5ZWU2MWQzYiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbm9uZS8zMmciLCJtb2RlbCI6ImhmLmNvL3Vuc2xvdGgvZ2VtbWEtNC1FMkItaXQtR0dVRjpJUTRfWFMiLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6MzYsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6MTExNS43MjA1MzcwMDAwMDA0LCJ0b3RhbF90b2tlbnMiOjc3MzJ9fQ==",
            "payload_sha256": "nuEAkZfrnuT21JxTIe560JywU1vX1SKGrJF0SwwTVyU="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-iq4_xs-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Windows PC with Intel graphics, 16 GB",
          "machine_detail": null,
          "machine_class": "windows/amd64/intel/16g",
          "machines": 1,
          "tokens_per_sec": 6.8,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 10.9041,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates",
          "measured_at": "2026-09-14T02:26:48.645Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38"
          ]
        },
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": 6.8,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
          "engine": "ollama",
          "cost_usd_per_million": 10.9087,
          "cost_breakdown": {
            "tokens_per_sec": 6.8,
            "usd_per_hour": 0.2669,
            "usd_per_million": 10.9087,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 10.2986
              },
              {
                "resource": "Memory",
                "quantity": "3.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.175,
                "usd_per_million": 0.4671
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.143
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "BMywkb7dV7N0NCPCAwy2frRvD+1b5KnYKmhCfTLOR3UmayTD4dvFcTYzmq6G85h+OCYe1P+rzHd/GFQsWLZOBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "2BWBi/8yZXNYa1whXNPsNiCfdhfLDrd6wTyXZiJQjI8="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 5.33,
          "context_tokens": 4096,
          "build": "Q8_0 (8-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 6.4387,
          "cost_breakdown": {
            "tokens_per_sec": 5.33,
            "usd_per_hour": 0.1236,
            "usd_per_million": 6.4387,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 5.6258
              },
              {
                "resource": "Memory",
                "quantity": "4.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.918,
                "usd_per_million": 0.7347
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0782
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "lag0lwjS4kyWzwpRkRHVnEMNQH0ca4joB4rweunr9B64XjrGwSOY0rioLtiDVxRY3fIyXtD8bDGVeHp4MNKODg==",
            "signing_payload_b64": "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",
            "payload_sha256": "JiwOrHWqI9pNus/avHdZrof1LjcUpdDMap6yHsAhQDg="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q8_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Intel Mac, 8 GB",
          "machine_detail": null,
          "machine_class": "darwin/amd64/apple/8g",
          "machines": 1,
          "tokens_per_sec": 3.99,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 8.4191,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates",
          "measured_at": "2026-10-22T10:43:12.498Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W43"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 3.14,
          "context_tokens": 4096,
          "build": "BF16 (16-bit, full precision) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:BF16",
          "engine": "ollama-host",
          "cost_usd_per_million": 11.9727,
          "cost_breakdown": {
            "tokens_per_sec": 3.14,
            "usd_per_hour": 0.1355,
            "usd_per_million": 11.9727,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 9.5418
              },
              {
                "resource": "Memory",
                "quantity": "8.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 7.226,
                "usd_per_million": 2.2983
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1326
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "R55I6Zr0hGc5e3kdiZi4j5CevKAvxSkvg3dMzkXrrnvrBMXcRB8e/nVe6GMymbJUmV3ZZd+PUp4xDjc0JXuOCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "++4DHPgbc72KmX2S0OxSXVVxNb4R159M4IXYxqMSBC4="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-bf16-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Mac · no GPU · 4 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 4 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 12.7.6",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|amd64|4|8|none|0|||"
          },
          "machine_class": "darwin|amd64|4|8|none|0|||",
          "machines": 1,
          "tokens_per_sec": 2.78,
          "context_tokens": 8192,
          "build": "Q3_K_M (3-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q3_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 11.5908,
          "cost_breakdown": {
            "tokens_per_sec": 2.78,
            "usd_per_hour": 0.1159,
            "usd_per_million": 11.5908,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 10.8009
              },
              {
                "resource": "Memory",
                "quantity": "2.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.777,
                "usd_per_million": 0.6398
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1501
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Core i5-5350U",
              "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
              "clock_ghz": 1.8,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 3,
              "single_core_hashes_per_sec": 93.7,
              "multi_core_hashes_per_sec": 195.1,
              "sustained_ratio": 0.7,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T23:57:35.534Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 8,
              "bandwidth_gb_s": 0.1,
              "latency_ns": 170.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T23:57:37.012Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 89.9,
              "write_mb_s": 44.3,
              "random_read_iops": 25045,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T23:57:40.536Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-16T00:27:19.106Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "PQIevpHHIJq7FIIU/zB39melUPlvBe6fyOs5bNdTiquVHiYQx5m6tudJxXrsqtgFW1+RL6qDCPX30rY+E9rRAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "BC/6GOLalZ0BS2F3c5ndM0SLFG03Umqq547zJo6sgns="
          },
          "build_page": "/inferences/gemma-4/e2b/unsloth-q3_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W34",
            "2026-W35"
          ],
          "providers": [
            "d68672"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "BF16 (16-bit, full precision) by bartowski",
          "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:BF16",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T14:27:05.533Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e2b~bartowski/bartowski-bf16-ollama",
          "hub_url": "https://huggingface.co/bartowski/google_gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "IQ2_M (2-bit) by bartowski",
          "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:IQ2_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T16:08:01.969Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e2b~bartowski/bartowski-iq2_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/google_gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "IQ3_XS (3-bit) by bartowski",
          "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:IQ3_XS",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T16:16:48.547Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e2b~bartowski/bartowski-iq3_xs-ollama",
          "hub_url": "https://huggingface.co/bartowski/google_gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q3_K_L (3-bit) by bartowski",
          "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T16:24:05.055Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e2b~bartowski/bartowski-q3_k_l-ollama",
          "hub_url": "https://huggingface.co/bartowski/google_gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q4_0 (4-bit) by bartowski",
          "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T16:31:33.170Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e2b~bartowski/bartowski-q4_0-ollama",
          "hub_url": "https://huggingface.co/bartowski/google_gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T16:38:22.890Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e2b~bartowski/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/google_gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q8_0 (8-bit) by bartowski",
          "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T16:46:09.915Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e2b~bartowski/bartowski-q8_0-ollama",
          "hub_url": "https://huggingface.co/bartowski/google_gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "BF16 (16-bit, full precision) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:BF16",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-24T14:39:11.148Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e2b/unsloth-bf16-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q8_0 (8-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-24T18:02:16.226Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e2b/unsloth-q8_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/google/gemma-4-E2B-it"
      },
      "builds": 19,
      "variants": [],
      "bottom_line": "291 tokens a second on a Linux PC with an NVIDIA GPU, 256 GB. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Dirk-Qwen3.8-27B-GGUF",
      "made_by": "peculiar-ragdoll",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": "Qwen 3.8 27B",
      "variant_of": "/inferences/qwen-3-8/27b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/qwen-3-8/27b~peculiar-ragdoll-dirk",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "download_gb": 15.3,
      "params_b": 27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "machine not reported",
        "machine_detail": null,
        "machine_class": "unknown",
        "machines": 1,
        "tokens_per_sec": 9.96,
        "context_tokens": 131072,
        "build": "Q4_K_M (4-bit) by peculiar-ragdoll",
        "ref": "hf.co/peculiar-ragdoll/Dirk-Qwen3.8-27B-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "pQ4N6+DceuMzsSzHoKxRJ6/fEiCVCr4YZRPe1zSudDD1ULwYrANE2dFD8lCkFNPUJxenj18+icYYFbrgs0V2CA==",
          "signing_payload_b64": "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",
          "payload_sha256": "VffiTnQel940rVCn8SyIKzcmTb/d0Ski2syMkcnRrjY="
        },
        "build_page": "/inferences/qwen-3-8/27b~peculiar-ragdoll-dirk/peculiar-ragdoll-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/peculiar-ragdoll/dirk-qwen3.8-27b-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "machine not reported",
          "machine_detail": null,
          "machine_class": "unknown",
          "machines": 1,
          "tokens_per_sec": 9.96,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by peculiar-ragdoll",
          "ref": "hf.co/peculiar-ragdoll/Dirk-Qwen3.8-27B-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "pQ4N6+DceuMzsSzHoKxRJ6/fEiCVCr4YZRPe1zSudDD1ULwYrANE2dFD8lCkFNPUJxenj18+icYYFbrgs0V2CA==",
            "signing_payload_b64": "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",
            "payload_sha256": "VffiTnQel940rVCn8SyIKzcmTb/d0Ski2syMkcnRrjY="
          },
          "build_page": "/inferences/qwen-3-8/27b~peculiar-ragdoll-dirk/peculiar-ragdoll-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/peculiar-ragdoll/dirk-qwen3.8-27b-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.8-27b",
        "url": "https://openrouter.ai/qwen/qwen3.8-27b",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 1.875,
        "free_tier": false,
        "context_tokens": 1024000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.11,
            "usd_per_m_out": 0.781,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake",
            "quantization": null,
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 1.05,
            "context_length": 1024000,
            "uptime_1d": 97.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.039,
            "usd_per_m_out": 1.25,
            "context_length": 262144,
            "uptime_1d": 91.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 1.35,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cerebras",
            "tag": "cerebras/fp16",
            "quantization": "fp16",
            "usd_per_m_in": 0.99,
            "usd_per_m_out": 1.49,
            "context_length": 65536,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.035,
            "usd_per_m_out": 1.5,
            "context_length": 262144,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Tenstorrent",
            "tag": "tenstorrent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.6,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.875,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 1.95,
            "context_length": 262144,
            "uptime_1d": 91.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 1.98,
            "context_length": 262144,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.125,
            "usd_per_m_out": 2,
            "context_length": 262144,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 2.2,
            "context_length": 262144,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 2.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 2.3,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.425,
            "usd_per_m_out": 2.55,
            "context_length": 1000000,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.4,
            "usd_per_m_out": 3,
            "context_length": 262144,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.42,
            "usd_per_m_out": 3,
            "context_length": 1000000,
            "uptime_1d": 79.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 95.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 23
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.8-27B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "10.0 tokens a second on a machine not reported. OpenRouter sells it at $1.88 per million output tokens."
    },
    {
      "model": "Gemma 4 12B",
      "made_by": "Google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/gemma-4/12b",
      "hub_url": "https://huggingface.co/google/gemma-4-12B-it",
      "download_gb": 6.6,
      "params_b": 12,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 262144,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu",
        "Amharic",
        "Dutch",
        "Russian",
        "Swahili"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": 131072,
        "build": "Q4_0 (4-bit) by google",
        "ref": "hf.co/google/gemma-4-12B-it-qat-q4_0-gguf:Q4_0",
        "engine": "ollama-host",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-23T16:00:50.658Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/gemma-4/12b/google-q4_0-ollama",
        "hub_url": "https://huggingface.co/google/gemma-4-12b-it-qat-q4_0-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 5,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q4_0 (4-bit) by google",
          "ref": "hf.co/google/gemma-4-12B-it-qat-q4_0-gguf:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T16:00:50.658Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/12b/google-q4_0-ollama",
          "hub_url": "https://huggingface.co/google/gemma-4-12b-it-qat-q4_0-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "BF16 (16-bit, full precision) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-12b-it-GGUF:BF16",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-24T08:48:51.145Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/12b/unsloth-bf16-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-12b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "IQ4_XS (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-12b-it-GGUF:IQ4_XS",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T15:47:04.536Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/12b/unsloth-iq4_xs-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-12b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-12b-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-24T16:11:48.110Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/12b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-12b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 262144,
          "build": "Q8_0 (8-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-12b-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-24T17:37:21.280Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/12b/unsloth-q8_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-12b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/google/gemma-4-12B-it-qat-q4_0-unquantized"
      },
      "builds": 5,
      "variants": [],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded."
    },
    {
      "model": "Gemma 4 26B-A4B",
      "made_by": "Google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/gemma-4/26b-a4b",
      "hub_url": "https://huggingface.co/google/gemma-4-26B-A4B-it",
      "download_gb": 15.8,
      "params_b": 26,
      "active_params_b": 4,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Amharic",
        "Assamese",
        "Bengali",
        "English",
        "Spanish",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Nepali",
        "Dutch",
        "Odia",
        "Punjabi",
        "Russian",
        "Sinhala",
        "Swahili",
        "Tamil",
        "Telugu",
        "Urdu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": 131072,
        "build": "Q8_0 (8-bit) by unsloth",
        "ref": "hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:Q8_0",
        "engine": "ollama-host",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-24T13:57:16.814Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/gemma-4/26b-a4b/unsloth-q8_0-ollama",
        "hub_url": "https://huggingface.co/unsloth/gemma-4-26b-a4b-it-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q8_0 (8-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-24T13:57:16.814Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/26b-a4b/unsloth-q8_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-26b-a4b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "google/gemma-4-26b-a4b-it",
        "url": "https://openrouter.ai/google/gemma-4-26b-a4b-it",
        "input_usd_per_million": 0.0675,
        "output_usd_per_million": 0.225,
        "free_tier": true,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.017,
            "usd_per_m_out": 0.052,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom",
            "quantization": null,
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 0.15,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "NextBit",
            "tag": "nextbit/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.0675,
            "usd_per_m_out": 0.225,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.24,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora",
            "quantization": null,
            "usd_per_m_in": 0.09,
            "usd_per_m_out": 0.28,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Io Net",
            "tag": "io-net/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.067,
            "usd_per_m_out": 0.29,
            "context_length": 262142,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.3,
            "context_length": 262144,
            "uptime_1d": 93.1,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.3,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.33,
            "context_length": 262144,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.34,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.4,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.4,
            "context_length": 262144,
            "uptime_1d": 89.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.4,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Google",
            "tag": "google-vertex/global",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 262144,
            "uptime_1d": 91.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 15
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/google/gemma-4-26B-A4B-it"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded. OpenRouter sells it at $0.23 per million output tokens."
    },
    {
      "model": "Gemma 4 31B",
      "made_by": "Google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62ecdc18b72a69615d6bd857/E4lkPz1TZNLzIFr_dR273.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/gemma-4/31b",
      "hub_url": "https://huggingface.co/unsloth/gemma-4-31B-it-GGUF",
      "download_gb": null,
      "params_b": 31,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu",
        "Amharic",
        "Assamese",
        "Spanish",
        "Nepali",
        "Dutch",
        "Odia",
        "Russian",
        "Sinhala",
        "Swahili",
        "Urdu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": 131072,
        "build": "Q4_0 (4-bit) by unsloth",
        "ref": "hf.co/unsloth/gemma-4-31B-it-GGUF:Q4_0",
        "engine": "ollama-host",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-23T04:09:16.960Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/gemma-4/31b/unsloth-q4_0-ollama",
        "hub_url": "https://huggingface.co/unsloth/gemma-4-31b-it-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q4_0 (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-31B-it-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T04:09:16.960Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/31b/unsloth-q4_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-31b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 65536,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-31B-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-24T12:57:51.781Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/31b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-31b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "google/gemma-4-31b-it",
        "url": "https://openrouter.ai/google/gemma-4-31b-it",
        "input_usd_per_million": 0.2,
        "output_usd_per_million": 0.4,
        "free_tier": true,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.34,
            "usd_per_m_out": 0.1,
            "context_length": 131072,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.16,
            "usd_per_m_out": 0.32,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 0.335,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "KT",
            "tag": "kt",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.34,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.34,
            "context_length": 262144,
            "uptime_1d": 92.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.36,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Chutes",
            "tag": "chutes/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.37,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.4,
            "context_length": 262144,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.4,
            "context_length": 262144,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.4,
            "context_length": 262144,
            "uptime_1d": 96.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "NextBit",
            "tag": "nextbit/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 0.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/turbo",
            "quantization": "fp4",
            "usd_per_m_in": 0.135,
            "usd_per_m_out": 0.51,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 0.6,
            "context_length": 262144,
            "uptime_1d": 99.1,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 1,
            "context_length": 262144,
            "uptime_1d": 96.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SambaNova",
            "tag": "sambanova",
            "quantization": null,
            "usd_per_m_in": 0.38,
            "usd_per_m_out": 1.15,
            "context_length": 131072,
            "uptime_1d": 80,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 18
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/google/gemma-4-31B-it"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded. OpenRouter sells it at $0.40 per million output tokens."
    },
    {
      "model": "Gemma 4 E4B",
      "made_by": "Google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/gemma-4/e4b",
      "hub_url": "https://huggingface.co/google/gemma-4-E4B-it",
      "download_gb": 4.6,
      "params_b": 8,
      "active_params_b": 4,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Amharic",
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Dutch",
        "Punjabi",
        "Russian",
        "Swahili",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": 131072,
        "build": "Q4_K_M (4-bit) by google",
        "ref": "gemma4:e4b",
        "engine": "ollama-host",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-24T16:31:18.583Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/gemma-4/e4b/google-q4_k_m-ollama",
        "hub_url": null,
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 6,
      "weeks_measured": [
        "2026-W29",
        "2026-W35",
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by google",
          "ref": "gemma4:e4b",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-24T16:31:18.583Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e4b/google-q4_k_m-ollama",
          "hub_url": null,
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "BF16 (16-bit, full precision) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E4B-it-GGUF:BF16",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-24T15:19:31.012Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e4b/unsloth-bf16-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e4b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "IQ4_XS (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E4B-it-GGUF:IQ4_XS",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T15:22:20.859Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e4b/unsloth-iq4_xs-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e4b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q4_0 (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E4B-it-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T15:29:06.075Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e4b/unsloth-q4_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e4b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E4B-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T01:25:32.668Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e4b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e4b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q8_0 (8-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-4-E4B-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-24T14:52:59.345Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e4b/unsloth-q8_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e4b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/google/gemma-4-E4B-it"
      },
      "builds": 8,
      "variants": [],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded."
    },
    {
      "model": "MiniCPM-V 4.6 Base",
      "made_by": "openbmb",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1670387859384-633fe7784b362488336bbfad.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/minicpm-v-4-6/base",
      "hub_url": "https://huggingface.co/openbmb/MiniCPM-V-4.6",
      "download_gb": 0.5,
      "params_b": 0.87,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": 262144,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 130.91,
        "context_tokens": 262144,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM-V-4.6-gguf:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 1.6508,
        "cost_breakdown": {
          "tokens_per_sec": 130.91,
          "usd_per_hour": 0.778,
          "usd_per_million": 1.6508,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 0.8403
            },
            {
              "resource": "Memory",
              "quantity": "0.5 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.41,
              "usd_per_million": 0.0031
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.7639
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0435
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "yk9bevepOPKKjbVBVlsSlq4Jx5v0/2eP+Sx4OB43cug8U6GZ+3nlK40HK8jCrqoP91Fr+LIARO9kqVy2C4pFDQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "rmI44TzVINHmIPYTKm1EB2JYyrW2u5M1fRde5MAMkWA="
        },
        "build_page": "/inferences/minicpm-v-4-6/base/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm-v-4.6-gguf",
        "weeks": [
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 130.91,
        "context_tokens": 262144,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM-V-4.6-gguf:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 1.6508,
        "cost_breakdown": {
          "tokens_per_sec": 130.91,
          "usd_per_hour": 0.778,
          "usd_per_million": 1.6508,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 0.8403
            },
            {
              "resource": "Memory",
              "quantity": "0.5 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.41,
              "usd_per_million": 0.0031
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.7639
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0435
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "yk9bevepOPKKjbVBVlsSlq4Jx5v0/2eP+Sx4OB43cug8U6GZ+3nlK40HK8jCrqoP91Fr+LIARO9kqVy2C4pFDQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "rmI44TzVINHmIPYTKm1EB2JYyrW2u5M1fRde5MAMkWA="
        },
        "build_page": "/inferences/minicpm-v-4-6/base/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm-v-4.6-gguf",
        "weeks": [
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 2,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 130.91,
          "context_tokens": 262144,
          "build": "Q4_K_M (4-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM-V-4.6-gguf:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.6508,
          "cost_breakdown": {
            "tokens_per_sec": 130.91,
            "usd_per_hour": 0.778,
            "usd_per_million": 1.6508,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 0.8403
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.41,
                "usd_per_million": 0.0031
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.7639
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0435
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "yk9bevepOPKKjbVBVlsSlq4Jx5v0/2eP+Sx4OB43cug8U6GZ+3nlK40HK8jCrqoP91Fr+LIARO9kqVy2C4pFDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "rmI44TzVINHmIPYTKm1EB2JYyrW2u5M1fRde5MAMkWA="
          },
          "build_page": "/inferences/minicpm-v-4-6/base/openbmb-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm-v-4.6-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 27.29,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM-V-4.6-gguf:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 2.6154,
          "cost_breakdown": {
            "tokens_per_sec": 27.29,
            "usd_per_hour": 0.257,
            "usd_per_million": 2.6154,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 2.5648
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.41,
                "usd_per_million": 0.015
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0356
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "ZM26p9MeA5fgyWiG7k4BdWSDBgnPv+UxTwnE58tCPg5qKf9ug/s9VIrYNdEp+sjQf2xO03K48UX7+krIw9y6CQ==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MjcuMjkyODM5NDU5MTY4NTEsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0NzUxNDQ2ODgyNTg0OTAwLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwMTUzNmRmOTRmODhjMmYyZGNlOWM4NzI5N2Y0MmUwOTAyZDVkNjM1ZWEyNzM2Zjk5MTQxYzNlYmNlYTJlNWMwMiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoid2luZG93cy9hbWQ2NC9pbnRlbC8xNmciLCJtb2RlbCI6ImhmLmNvL29wZW5ibWIvTWluaUNQTS1WLTQuNi1nZ3VmOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYSIsInNhbXBsZV9jb3VudCI6NDIsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6NzE1LjIwNTkwNjk5OTk5OTksInRvdGFsX3Rva2VucyI6MTk1MjB9fQ==",
            "payload_sha256": "eOoDCzjxkJBjVnKripjXgZuVBgyAVx+9wyhhmja5ID8="
          },
          "build_page": "/inferences/minicpm-v-4-6/base/openbmb-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm-v-4.6-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openbmb/MiniCPM-V-4.6"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "131 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Huihui-Qwen3.8-27B-abliterated-GGUF",
      "made_by": "huihui-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/66228dc4c8920ec3513dc81a/9stczijVqp6TgQduzT8kN.png",
      "based_on": "Qwen 3.8 27B",
      "variant_of": "/inferences/qwen-3-8/27b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/qwen-3-8/27b~huihui-ai-abliterated",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "download_gb": 15.4,
      "params_b": 27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "machine not reported",
        "machine_detail": null,
        "machine_class": "unknown",
        "machines": 1,
        "tokens_per_sec": 11.84,
        "context_tokens": 131072,
        "build": "Q4_K_S (4-bit) by huihui-ai",
        "ref": "hf.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF:Q4_K_S",
        "engine": "ollama",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "X4KqT2b8Ux+rekeKsO/Qed9hq7p+oXl6si2jkg+Bn5OOiYqa5PHjgwPSF/SbsmP/mKfmxGW1SyE4ogawd+sEDg==",
          "signing_payload_b64": "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",
          "payload_sha256": "GncILOtDomyzVF4GcI3EcLZRWDxzwYHObfI2tWnvYzM="
        },
        "build_page": "/inferences/qwen-3-8/27b~huihui-ai-abliterated/huihui-ai-q4_k_s-ollama",
        "hub_url": "https://huggingface.co/huihui-ai/huihui-qwen3.8-27b-abliterated-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "machine not reported",
          "machine_detail": null,
          "machine_class": "unknown",
          "machines": 1,
          "tokens_per_sec": 11.84,
          "context_tokens": 131072,
          "build": "Q4_K_S (4-bit) by huihui-ai",
          "ref": "hf.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF:Q4_K_S",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "X4KqT2b8Ux+rekeKsO/Qed9hq7p+oXl6si2jkg+Bn5OOiYqa5PHjgwPSF/SbsmP/mKfmxGW1SyE4ogawd+sEDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "GncILOtDomyzVF4GcI3EcLZRWDxzwYHObfI2tWnvYzM="
          },
          "build_page": "/inferences/qwen-3-8/27b~huihui-ai-abliterated/huihui-ai-q4_k_s-ollama",
          "hub_url": "https://huggingface.co/huihui-ai/huihui-qwen3.8-27b-abliterated-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "machine not reported",
          "machine_detail": null,
          "machine_class": "unknown",
          "machines": 1,
          "tokens_per_sec": 9.35,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by huihui-ai",
          "ref": "hf.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "IdYd1QDlrgDR6jdw2xzbkM/916tQSNPJNokzK4kRdANU6f+ZwfFa3Zw7d5/uIQG1RRMQ9jnMARCh5rWrmL/JDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "J9OZyIhKBej42Qn/uik06CN2eFL9tIgTf4IPGsMlDbs="
          },
          "build_page": "/inferences/qwen-3-8/27b~huihui-ai-abliterated/huihui-ai-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/huihui-ai/huihui-qwen3.8-27b-abliterated-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.8-27b",
        "url": "https://openrouter.ai/qwen/qwen3.8-27b",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 1.875,
        "free_tier": false,
        "context_tokens": 1024000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.11,
            "usd_per_m_out": 0.781,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake",
            "quantization": null,
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 1.05,
            "context_length": 1024000,
            "uptime_1d": 97.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.039,
            "usd_per_m_out": 1.25,
            "context_length": 262144,
            "uptime_1d": 91.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 1.35,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cerebras",
            "tag": "cerebras/fp16",
            "quantization": "fp16",
            "usd_per_m_in": 0.99,
            "usd_per_m_out": 1.49,
            "context_length": 65536,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.035,
            "usd_per_m_out": 1.5,
            "context_length": 262144,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Tenstorrent",
            "tag": "tenstorrent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.6,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.875,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 1.95,
            "context_length": 262144,
            "uptime_1d": 91.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 1.98,
            "context_length": 262144,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.125,
            "usd_per_m_out": 2,
            "context_length": 262144,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 2.2,
            "context_length": 262144,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 2.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 2.3,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.425,
            "usd_per_m_out": 2.55,
            "context_length": 1000000,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.4,
            "usd_per_m_out": 3,
            "context_length": 262144,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.42,
            "usd_per_m_out": 3,
            "context_length": 1000000,
            "uptime_1d": 79.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 95.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 23
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.8-27B"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "12 tokens a second on a machine not reported. OpenRouter sells it at $1.88 per million output tokens."
    },
    {
      "model": "MiniCPM-V 4.5 Base",
      "made_by": "openbmb",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1670387859384-633fe7784b362488336bbfad.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/minicpm-v-4-5/base",
      "hub_url": "https://huggingface.co/openbmb/MiniCPM-V-4_5",
      "download_gb": 4.7,
      "params_b": 8.3,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": 16384,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 11.32,
        "context_tokens": 16384,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM-V-4_5-gguf:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 19.403,
        "cost_breakdown": {
          "tokens_per_sec": 11.32,
          "usd_per_hour": 0.7905,
          "usd_per_million": 19.403,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 9.7194
            },
            {
              "resource": "Memory",
              "quantity": "4.7 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 3.901,
              "usd_per_million": 0.3447
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 8.8358
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.5031
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "o3wXUzdllDEGjunbws96CoEA+QpKgELnVgd1uyT4SNCcT8Hd5DBRPua9GjFy/f0t2fH7VI0reYINkoj5ootXCQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "0lOoDd82ljreZYZRB/Vhquv8nK+wM2Jwuu1gPoyXKYM="
        },
        "build_page": "/inferences/minicpm-v-4-5/base/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm-v-4_5-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 11.32,
        "context_tokens": 16384,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM-V-4_5-gguf:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 19.403,
        "cost_breakdown": {
          "tokens_per_sec": 11.32,
          "usd_per_hour": 0.7905,
          "usd_per_million": 19.403,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 9.7194
            },
            {
              "resource": "Memory",
              "quantity": "4.7 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 3.901,
              "usd_per_million": 0.3447
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 8.8358
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.5031
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "o3wXUzdllDEGjunbws96CoEA+QpKgELnVgd1uyT4SNCcT8Hd5DBRPua9GjFy/f0t2fH7VI0reYINkoj5ootXCQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "0lOoDd82ljreZYZRB/Vhquv8nK+wM2Jwuu1gPoyXKYM="
        },
        "build_page": "/inferences/minicpm-v-4-5/base/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm-v-4_5-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 11.32,
          "context_tokens": 16384,
          "build": "Q4_K_M (4-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM-V-4_5-gguf:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 19.403,
          "cost_breakdown": {
            "tokens_per_sec": 11.32,
            "usd_per_hour": 0.7905,
            "usd_per_million": 19.403,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 9.7194
              },
              {
                "resource": "Memory",
                "quantity": "4.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.901,
                "usd_per_million": 0.3447
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 8.8358
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.5031
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "o3wXUzdllDEGjunbws96CoEA+QpKgELnVgd1uyT4SNCcT8Hd5DBRPua9GjFy/f0t2fH7VI0reYINkoj5ootXCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "0lOoDd82ljreZYZRB/Vhquv8nK+wM2Jwuu1gPoyXKYM="
          },
          "build_page": "/inferences/minicpm-v-4-5/base/openbmb-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm-v-4_5-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openbmb/MiniCPM-V-4_5"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "11 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Qwen3.8-27B-Uncensored-GGUF",
      "made_by": "JonathanColetti",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": "Qwen 3.8 27B",
      "variant_of": "/inferences/qwen-3-8/27b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/qwen-3-8/27b~jonathancoletti-uncensored",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "download_gb": 15.7,
      "params_b": 27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "machine not reported",
        "machine_detail": null,
        "machine_class": "unknown",
        "machines": 1,
        "tokens_per_sec": 9.85,
        "context_tokens": 131072,
        "build": "Q4_K_M (4-bit) by JonathanColetti",
        "ref": "hf.co/JonathanColetti/Qwen3.8-27B-Uncensored-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "l1UIN1vv+4GJrhT006hy9Z5HYchMPUEYPmEIJ+wM77WD9lSUcZym2OSLsM1WR6B4wjM4dul3SkSnDLLUbOUpAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "mspraICjmgmzpLjarzj2YZZUFQAd97XDm10dEFTVb+4="
        },
        "build_page": "/inferences/qwen-3-8/27b~jonathancoletti-uncensored/jonathancoletti-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/jonathancoletti/qwen3.8-27b-uncensored-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "machine not reported",
          "machine_detail": null,
          "machine_class": "unknown",
          "machines": 1,
          "tokens_per_sec": 9.85,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by JonathanColetti",
          "ref": "hf.co/JonathanColetti/Qwen3.8-27B-Uncensored-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "l1UIN1vv+4GJrhT006hy9Z5HYchMPUEYPmEIJ+wM77WD9lSUcZym2OSLsM1WR6B4wjM4dul3SkSnDLLUbOUpAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "mspraICjmgmzpLjarzj2YZZUFQAd97XDm10dEFTVb+4="
          },
          "build_page": "/inferences/qwen-3-8/27b~jonathancoletti-uncensored/jonathancoletti-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/jonathancoletti/qwen3.8-27b-uncensored-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "machine not reported",
          "machine_detail": null,
          "machine_class": "unknown",
          "machines": 1,
          "tokens_per_sec": 3.37,
          "context_tokens": 131072,
          "build": "Q5_K_M (5-bit) by JonathanColetti",
          "ref": "hf.co/JonathanColetti/Qwen3.8-27B-Uncensored-GGUF:Q5_K_M",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "AL5e4Zc/CHAs2xEAIF2X+oVXoRzay10SMnbcme5quL/5kJZeS6o/k/eoHsTxTlIcM18q9YSFlereiMRCWSLVBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "Uu9bCdn9pUcPbGtiZOQLolmNDP5lIHjFNM8j35mjzfE="
          },
          "build_page": "/inferences/qwen-3-8/27b~jonathancoletti-uncensored/jonathancoletti-q5_k_m-ollama",
          "hub_url": "https://huggingface.co/jonathancoletti/qwen3.8-27b-uncensored-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.8-27b",
        "url": "https://openrouter.ai/qwen/qwen3.8-27b",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 1.875,
        "free_tier": false,
        "context_tokens": 1024000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.11,
            "usd_per_m_out": 0.781,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake",
            "quantization": null,
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 1.05,
            "context_length": 1024000,
            "uptime_1d": 97.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.039,
            "usd_per_m_out": 1.25,
            "context_length": 262144,
            "uptime_1d": 91.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 1.35,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cerebras",
            "tag": "cerebras/fp16",
            "quantization": "fp16",
            "usd_per_m_in": 0.99,
            "usd_per_m_out": 1.49,
            "context_length": 65536,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.035,
            "usd_per_m_out": 1.5,
            "context_length": 262144,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Tenstorrent",
            "tag": "tenstorrent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.6,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.875,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 1.95,
            "context_length": 262144,
            "uptime_1d": 91.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 1.98,
            "context_length": 262144,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.125,
            "usd_per_m_out": 2,
            "context_length": 262144,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 2.2,
            "context_length": 262144,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 2.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 2.3,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.425,
            "usd_per_m_out": 2.55,
            "context_length": 1000000,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.4,
            "usd_per_m_out": 3,
            "context_length": 262144,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.42,
            "usd_per_m_out": 3,
            "context_length": 1000000,
            "uptime_1d": 79.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 95.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 23
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.8-27B"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "9.8 tokens a second on a machine not reported. OpenRouter sells it at $1.88 per million output tokens."
    },
    {
      "model": "Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF",
      "made_by": "huihui-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/66228dc4c8920ec3513dc81a/9stczijVqp6TgQduzT8kN.png",
      "based_on": "Qwen 3.6 35B-A3B",
      "variant_of": "/inferences/qwen-3-6/35b-a3b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/qwen-3-6/35b-a3b~huihui-ai-abliterated-mtp",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B",
      "download_gb": 12.3,
      "params_b": 35,
      "active_params_b": 3,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Tamil",
        "Amharic",
        "Assamese",
        "Spanish",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Nepali",
        "Dutch",
        "Punjabi",
        "Russian",
        "Sinhala",
        "Telugu",
        "Urdu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 5,
        "context_tokens": null,
        "build": "Q2_K (2-bit) by huihui-ai",
        "ref": "hf.co/huihui-ai/Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF:Q2_K",
        "engine": "ollama-host",
        "cost_usd_per_million": 8.1457,
        "cost_breakdown": {
          "tokens_per_sec": 5,
          "usd_per_hour": 0.1465,
          "usd_per_million": 8.1457,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 6.0046
            },
            {
              "resource": "Memory",
              "quantity": "12.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 10.28,
              "usd_per_million": 2.0576
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0835
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 185.9,
            "multi_core_hashes_per_sec": 588,
            "sustained_ratio": 1.4,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-14T19:37:12.663Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 1.7,
            "latency_ns": 71.3,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-14T19:37:15.832Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2202.4,
            "write_mb_s": 15.5,
            "random_read_iops": 196685,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-14T19:37:24.078Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-14T20:22:46.948Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
          "signature": "Qx6tfy6v914JGxzCFocDcsVZ5t/8lZ6vavl+giiW31Mq+vVESnSznpVelFwDxUXg2FW/vxg7uGPTeyoVdGntBw==",
          "signing_payload_b64": "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",
          "payload_sha256": "j3kCJ/BpEf4p9Wx3upEBKBigpvY0nO8INwQlVebUDeU="
        },
        "build_page": "/inferences/qwen-3-6/35b-a3b~huihui-ai-abliterated-mtp/huihui-ai-q2_k-ollama",
        "hub_url": "https://huggingface.co/huihui-ai/huihui-qwen3.6-35b-a3b-abliterated-mtp-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "4e94b2"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 5,
        "context_tokens": null,
        "build": "Q2_K (2-bit) by huihui-ai",
        "ref": "hf.co/huihui-ai/Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF:Q2_K",
        "engine": "ollama-host",
        "cost_usd_per_million": 8.1457,
        "cost_breakdown": {
          "tokens_per_sec": 5,
          "usd_per_hour": 0.1465,
          "usd_per_million": 8.1457,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 6.0046
            },
            {
              "resource": "Memory",
              "quantity": "12.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 10.28,
              "usd_per_million": 2.0576
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0835
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 185.9,
            "multi_core_hashes_per_sec": 588,
            "sustained_ratio": 1.4,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-14T19:37:12.663Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 1.7,
            "latency_ns": 71.3,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-14T19:37:15.832Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2202.4,
            "write_mb_s": 15.5,
            "random_read_iops": 196685,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-14T19:37:24.078Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-14T20:22:46.948Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
          "signature": "Qx6tfy6v914JGxzCFocDcsVZ5t/8lZ6vavl+giiW31Mq+vVESnSznpVelFwDxUXg2FW/vxg7uGPTeyoVdGntBw==",
          "signing_payload_b64": "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",
          "payload_sha256": "j3kCJ/BpEf4p9Wx3upEBKBigpvY0nO8INwQlVebUDeU="
        },
        "build_page": "/inferences/qwen-3-6/35b-a3b~huihui-ai-abliterated-mtp/huihui-ai-q2_k-ollama",
        "hub_url": "https://huggingface.co/huihui-ai/huihui-qwen3.6-35b-a3b-abliterated-mtp-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "4e94b2"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W44",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 5,
          "context_tokens": null,
          "build": "Q2_K (2-bit) by huihui-ai",
          "ref": "hf.co/huihui-ai/Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF:Q2_K",
          "engine": "ollama-host",
          "cost_usd_per_million": 8.1457,
          "cost_breakdown": {
            "tokens_per_sec": 5,
            "usd_per_hour": 0.1465,
            "usd_per_million": 8.1457,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.0046
              },
              {
                "resource": "Memory",
                "quantity": "12.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 10.28,
                "usd_per_million": 2.0576
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0835
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 185.9,
              "multi_core_hashes_per_sec": 588,
              "sustained_ratio": 1.4,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-14T19:37:12.663Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 1.7,
              "latency_ns": 71.3,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-14T19:37:15.832Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2202.4,
              "write_mb_s": 15.5,
              "random_read_iops": 196685,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-14T19:37:24.078Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T20:22:46.948Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "Qx6tfy6v914JGxzCFocDcsVZ5t/8lZ6vavl+giiW31Mq+vVESnSznpVelFwDxUXg2FW/vxg7uGPTeyoVdGntBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "j3kCJ/BpEf4p9Wx3upEBKBigpvY0nO8INwQlVebUDeU="
          },
          "build_page": "/inferences/qwen-3-6/35b-a3b~huihui-ai-abliterated-mtp/huihui-ai-q2_k-ollama",
          "hub_url": "https://huggingface.co/huihui-ai/huihui-qwen3.6-35b-a3b-abliterated-mtp-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "4e94b2"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 2,
          "tokens_per_sec": 4.29,
          "context_tokens": 4096,
          "build": "Q6_K (6-bit) by huihui-ai",
          "ref": "hf.co/huihui-ai/Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF:Q6_K",
          "engine": "ollama-host",
          "cost_usd_per_million": 12.3819,
          "cost_breakdown": {
            "tokens_per_sec": 4.29,
            "usd_per_hour": 0.1911,
            "usd_per_million": 12.3819,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.9974
              },
              {
                "resource": "Memory",
                "quantity": "27.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 22.668,
                "usd_per_million": 5.2872
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0973
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 185.9,
              "multi_core_hashes_per_sec": 588,
              "sustained_ratio": 1.4,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-14T19:37:12.663Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 1.7,
              "latency_ns": 71.3,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-14T19:37:15.832Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2202.4,
              "write_mb_s": 15.5,
              "random_read_iops": 196685,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-14T19:37:24.078Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T20:22:46.948Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "xt9ES5v2myeNju0C0/oQI786YQUEMxKxNpPpbPmWpzzTRJnOHbS7qby1zi/guxzZNOh+j2ad+ED3z+Y8pOVPCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "A7asW5Anl42qeNaOTyfVt5iKgX98VQhbPu/Ahb7aZ+k="
          },
          "build_page": "/inferences/qwen-3-6/35b-a3b~huihui-ai-abliterated-mtp/huihui-ai-q6_k-ollama",
          "hub_url": "https://huggingface.co/huihui-ai/huihui-qwen3.6-35b-a3b-abliterated-mtp-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "4e94b2",
            "e3654f"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.6-35b-a3b",
        "url": "https://openrouter.ai/qwen/qwen3.6-35b-a3b",
        "input_usd_per_million": 0.1,
        "output_usd_per_million": 0.9,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.066,
            "usd_per_m_out": 0.602,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.05,
            "usd_per_m_out": 0.7,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.76,
            "context_length": 262144,
            "uptime_1d": 85.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.89,
            "context_length": 262144,
            "uptime_1d": 91.9,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.9,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.9,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.95,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 1,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 1,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.25,
            "usd_per_m_out": 1.25,
            "context_length": 262144,
            "uptime_1d": 29.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.425,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 1.8,
            "context_length": 262144,
            "uptime_1d": 95.3,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 12
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "5.0 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. About $8.15 per million tokens here, against OpenRouter's $0.90: the API is 9.1× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
      "made_by": "llmfan46",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68851b893b66feaa5ca027d5/pxwPI83JFkYGWh5zD2r_c.png",
      "based_on": "Qwen 3.6 35B-A3B",
      "variant_of": "/inferences/qwen-3-6/35b-a3b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/qwen-3-6/35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved",
      "hub_url": "https://huggingface.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
      "download_gb": 20.3,
      "params_b": 35,
      "active_params_b": 3,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu",
        "Gujarati"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 4.85,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by llmfan46",
        "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 9.7638,
        "cost_breakdown": {
          "tokens_per_sec": 4.85,
          "usd_per_hour": 0.1704,
          "usd_per_million": 9.7638,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 6.1901
            },
            {
              "resource": "Memory",
              "quantity": "20.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 16.903,
              "usd_per_million": 3.4877
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.086
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 185.9,
            "multi_core_hashes_per_sec": 588,
            "sustained_ratio": 1.4,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-14T19:37:12.663Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 1.7,
            "latency_ns": 71.3,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-14T19:37:15.832Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2202.4,
            "write_mb_s": 15.5,
            "random_read_iops": 196685,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-14T19:37:24.078Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-14T20:22:46.948Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
          "signature": "WN4iNJwRNUhu2+5s3dXJh22mY+r6fzai73ExjbgOV6czlBNQot7uqwyDDHAMHruHbcUoH2f6YN/lNmcDYkn7Cg==",
          "signing_payload_b64": "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",
          "payload_sha256": "lfXVJz9fTBsyK2mY2vgif704gHEGO90xq6bwKgAJ49c="
        },
        "build_page": "/inferences/qwen-3-6/35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved/llmfan46-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/llmfan46/qwen3.6-35b-a3b-uncensored-heretic-native-mtp-preserved-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "4e94b2"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 4.83,
        "context_tokens": 4096,
        "build": "Q3_K_M (3-bit) by llmfan46",
        "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q3_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 9.0744,
        "cost_breakdown": {
          "tokens_per_sec": 4.83,
          "usd_per_hour": 0.1577,
          "usd_per_million": 9.0744,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 6.2128
            },
            {
              "resource": "Memory",
              "quantity": "16.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 13.401,
              "usd_per_million": 2.7753
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0864
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 185.9,
            "multi_core_hashes_per_sec": 588,
            "sustained_ratio": 1.4,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-14T19:37:12.663Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 1.7,
            "latency_ns": 71.3,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-14T19:37:15.832Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2202.4,
            "write_mb_s": 15.5,
            "random_read_iops": 196685,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-14T19:37:24.078Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-14T20:22:46.948Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
          "signature": "Y1NbHgJQ/4PEDQeSrJuccaqwUYeKO/oPhq47dAyzBA2yJ32EoWadilE4V/BlnMetdw3Z9J4wV4sKhHDNV9erBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "RcN3JWbOSr5kRjFeICrWyn+O70FtCptdmZ8MjqsK3JE="
        },
        "build_page": "/inferences/qwen-3-6/35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved/llmfan46-q3_k_m-ollama",
        "hub_url": "https://huggingface.co/llmfan46/qwen3.6-35b-a3b-uncensored-heretic-native-mtp-preserved-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "4e94b2"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 4.85,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by llmfan46",
          "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 9.7638,
          "cost_breakdown": {
            "tokens_per_sec": 4.85,
            "usd_per_hour": 0.1704,
            "usd_per_million": 9.7638,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.1901
              },
              {
                "resource": "Memory",
                "quantity": "20.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 16.903,
                "usd_per_million": 3.4877
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.086
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 185.9,
              "multi_core_hashes_per_sec": 588,
              "sustained_ratio": 1.4,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-14T19:37:12.663Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 1.7,
              "latency_ns": 71.3,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-14T19:37:15.832Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2202.4,
              "write_mb_s": 15.5,
              "random_read_iops": 196685,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-14T19:37:24.078Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T20:22:46.948Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "WN4iNJwRNUhu2+5s3dXJh22mY+r6fzai73ExjbgOV6czlBNQot7uqwyDDHAMHruHbcUoH2f6YN/lNmcDYkn7Cg==",
            "signing_payload_b64": "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",
            "payload_sha256": "lfXVJz9fTBsyK2mY2vgif704gHEGO90xq6bwKgAJ49c="
          },
          "build_page": "/inferences/qwen-3-6/35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved/llmfan46-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/llmfan46/qwen3.6-35b-a3b-uncensored-heretic-native-mtp-preserved-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "4e94b2"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 4.83,
          "context_tokens": 4096,
          "build": "Q3_K_M (3-bit) by llmfan46",
          "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q3_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 9.0744,
          "cost_breakdown": {
            "tokens_per_sec": 4.83,
            "usd_per_hour": 0.1577,
            "usd_per_million": 9.0744,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.2128
              },
              {
                "resource": "Memory",
                "quantity": "16.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 13.401,
                "usd_per_million": 2.7753
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0864
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 185.9,
              "multi_core_hashes_per_sec": 588,
              "sustained_ratio": 1.4,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-14T19:37:12.663Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 1.7,
              "latency_ns": 71.3,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-14T19:37:15.832Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2202.4,
              "write_mb_s": 15.5,
              "random_read_iops": 196685,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-14T19:37:24.078Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T20:22:46.948Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "Y1NbHgJQ/4PEDQeSrJuccaqwUYeKO/oPhq47dAyzBA2yJ32EoWadilE4V/BlnMetdw3Z9J4wV4sKhHDNV9erBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "RcN3JWbOSr5kRjFeICrWyn+O70FtCptdmZ8MjqsK3JE="
          },
          "build_page": "/inferences/qwen-3-6/35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved/llmfan46-q3_k_m-ollama",
          "hub_url": "https://huggingface.co/llmfan46/qwen3.6-35b-a3b-uncensored-heretic-native-mtp-preserved-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "4e94b2"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 4.81,
          "context_tokens": 4096,
          "build": "Q4_K_S (4-bit) by llmfan46",
          "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q4_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 9.6125,
          "cost_breakdown": {
            "tokens_per_sec": 4.81,
            "usd_per_hour": 0.1666,
            "usd_per_million": 9.6125,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.2316
              },
              {
                "resource": "Memory",
                "quantity": "19.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 15.859,
                "usd_per_million": 3.2942
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0866
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 185.9,
              "multi_core_hashes_per_sec": 588,
              "sustained_ratio": 1.4,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-14T19:37:12.663Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 1.7,
              "latency_ns": 71.3,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-14T19:37:15.832Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2202.4,
              "write_mb_s": 15.5,
              "random_read_iops": 196685,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-14T19:37:24.078Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T20:22:46.948Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "1lVy2ogFSB1naP+mYMnc1y3T2J+Q3zyHtCO6+6FdAZ+L6t+kz7DcP9KeTUNFdY8ct2agePktB2B8jj9G/jCQAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "iCxNCfxqp9SUaRneojSyd6Te9RqIVXEBXHHsxabP7vY="
          },
          "build_page": "/inferences/qwen-3-6/35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved/llmfan46-q4_k_s-ollama",
          "hub_url": "https://huggingface.co/llmfan46/qwen3.6-35b-a3b-uncensored-heretic-native-mtp-preserved-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "4e94b2"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "4.8 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Qwen 3.8 27B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3-8/27b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "download_gb": 15,
      "params_b": 27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 8192,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Amharic",
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Dutch",
        "Punjabi",
        "Russian",
        "Swahili",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "Tesla T4",
          "gpu_count": 2,
          "vram_gb": 15,
          "host": "Kaggle",
          "vendor_org": "nvidia",
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
        "machines": 1,
        "tokens_per_sec": 2.77,
        "context_tokens": 4096,
        "build": "Q8_0 (8-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3.8-27B-GGUF:Q8_0",
        "engine": "ollama-host",
        "cost_usd_per_million": 94.304,
        "cost_breakdown": {
          "tokens_per_sec": 2.77,
          "usd_per_hour": 0.9407,
          "usd_per_million": 94.304,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 10.8274
            },
            {
              "resource": "Memory",
              "quantity": "27.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 22.543,
              "usd_per_million": 8.1361
            },
            {
              "resource": "GPU",
              "quantity": "2 of 2 cards (the cards this build is split across)",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 200,
              "usd_per_million": 72.1826
            },
            {
              "resource": "Energy",
              "quantity": "315 W modelled (3 threads × 5 W + 2 × 150 W cards)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 8.75,
              "usd_per_million": 3.158
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "NAgKl9Ub2Xiw2WykYiqFsP3ChrRz44sKSYXWIqB9klPPsBM6ElJDlpgWo3YWfcFIrhgIc5RXaZkprfOgk8f5Dw==",
          "signing_payload_b64": "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",
          "payload_sha256": "tZcwj+6t8SZuD7d484G+Pqse1jm8aakRqzhp0RIqDTo="
        },
        "build_page": "/inferences/qwen-3-8/27b/unsloth-q8_0-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3.8-27b-gguf",
        "weeks": [
          "2026-W44",
          "2026-W45"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": {
        "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "Tesla T4",
          "gpu_count": 2,
          "vram_gb": 15,
          "host": "Kaggle",
          "vendor_org": "nvidia",
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
        "machines": 1,
        "tokens_per_sec": 2.77,
        "context_tokens": 4096,
        "build": "Q8_0 (8-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3.8-27B-GGUF:Q8_0",
        "engine": "ollama-host",
        "cost_usd_per_million": 94.304,
        "cost_breakdown": {
          "tokens_per_sec": 2.77,
          "usd_per_hour": 0.9407,
          "usd_per_million": 94.304,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 10.8274
            },
            {
              "resource": "Memory",
              "quantity": "27.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 22.543,
              "usd_per_million": 8.1361
            },
            {
              "resource": "GPU",
              "quantity": "2 of 2 cards (the cards this build is split across)",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 200,
              "usd_per_million": 72.1826
            },
            {
              "resource": "Energy",
              "quantity": "315 W modelled (3 threads × 5 W + 2 × 150 W cards)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 8.75,
              "usd_per_million": 3.158
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "NAgKl9Ub2Xiw2WykYiqFsP3ChrRz44sKSYXWIqB9klPPsBM6ElJDlpgWo3YWfcFIrhgIc5RXaZkprfOgk8f5Dw==",
          "signing_payload_b64": "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",
          "payload_sha256": "tZcwj+6t8SZuD7d484G+Pqse1jm8aakRqzhp0RIqDTo="
        },
        "build_page": "/inferences/qwen-3-8/27b/unsloth-q8_0-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3.8-27b-gguf",
        "weeks": [
          "2026-W44",
          "2026-W45"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "machine_kinds": 2,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W44",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "Tesla T4",
            "gpu_count": 2,
            "vram_gb": 15,
            "host": "Kaggle",
            "vendor_org": "nvidia",
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
          "machines": 1,
          "tokens_per_sec": 2.77,
          "context_tokens": 4096,
          "build": "Q8_0 (8-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3.8-27B-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 94.304,
          "cost_breakdown": {
            "tokens_per_sec": 2.77,
            "usd_per_hour": 0.9407,
            "usd_per_million": 94.304,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 10.8274
              },
              {
                "resource": "Memory",
                "quantity": "27.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 22.543,
                "usd_per_million": 8.1361
              },
              {
                "resource": "GPU",
                "quantity": "2 of 2 cards (the cards this build is split across)",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 200,
                "usd_per_million": 72.1826
              },
              {
                "resource": "Energy",
                "quantity": "315 W modelled (3 threads × 5 W + 2 × 150 W cards)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 8.75,
                "usd_per_million": 3.158
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "NAgKl9Ub2Xiw2WykYiqFsP3ChrRz44sKSYXWIqB9klPPsBM6ElJDlpgWo3YWfcFIrhgIc5RXaZkprfOgk8f5Dw==",
            "signing_payload_b64": "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",
            "payload_sha256": "tZcwj+6t8SZuD7d484G+Pqse1jm8aakRqzhp0RIqDTo="
          },
          "build_page": "/inferences/qwen-3-8/27b/unsloth-q8_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.8-27b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "Tesla T4",
            "gpu_count": 2,
            "vram_gb": 15,
            "host": "Kaggle",
            "vendor_org": "nvidia",
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 8192,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "unsloth/Qwen3.8-27B-GGUF:Q4_K_M",
          "engine": "llamacpp",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 11.5,
              "watts": 105,
              "tflops_per_watt": 0.109,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-10-28T10:39:20.231Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-28T07:01:10.429Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-3-8/27b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.8-27b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "unsloth/Qwen3.8-27B-GGUF:Q4_K_M",
          "engine": "llamacpp",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T01:35:22.479Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-3-8/27b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.8-27b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.8-27b",
        "url": "https://openrouter.ai/qwen/qwen3.8-27b",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 1.875,
        "free_tier": false,
        "context_tokens": 1024000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.11,
            "usd_per_m_out": 0.781,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake",
            "quantization": null,
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 1.05,
            "context_length": 1024000,
            "uptime_1d": 97.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.039,
            "usd_per_m_out": 1.25,
            "context_length": 262144,
            "uptime_1d": 91.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 1.35,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cerebras",
            "tag": "cerebras/fp16",
            "quantization": "fp16",
            "usd_per_m_in": 0.99,
            "usd_per_m_out": 1.49,
            "context_length": 65536,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.035,
            "usd_per_m_out": 1.5,
            "context_length": 262144,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Tenstorrent",
            "tag": "tenstorrent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.6,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.875,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 1.95,
            "context_length": 262144,
            "uptime_1d": 91.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 1.98,
            "context_length": 262144,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.125,
            "usd_per_m_out": 2,
            "context_length": 262144,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 2.2,
            "context_length": 262144,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 2.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 2.3,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.425,
            "usd_per_m_out": 2.55,
            "context_length": 1000000,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.4,
            "usd_per_m_out": 3,
            "context_length": 262144,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.42,
            "usd_per_m_out": 3,
            "context_length": 1000000,
            "uptime_1d": 79.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 95.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 23
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.8-27B"
      },
      "builds": 7,
      "variants": [
        {
          "model": "Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP-GGUF",
          "made_by": "HauhauCS",
          "page": "/inferences/qwen-3-8/27b~hauhaucs-uncensored-aggressive-mtp",
          "status": "measured",
          "tokens_per_sec": 1.75
        },
        {
          "model": "Qwen3.8-27B-Uncensored-GGUF",
          "made_by": "JonathanColetti",
          "page": "/inferences/qwen-3-8/27b~jonathancoletti-uncensored",
          "status": "measured",
          "tokens_per_sec": 9.85
        },
        {
          "model": "Huihui-Qwen3.8-27B-abliterated-GGUF",
          "made_by": "huihui-ai",
          "page": "/inferences/qwen-3-8/27b~huihui-ai-abliterated",
          "status": "measured",
          "tokens_per_sec": 11.84
        },
        {
          "model": "Dirk-Qwen3.8-27B-GGUF",
          "made_by": "peculiar-ragdoll",
          "page": "/inferences/qwen-3-8/27b~peculiar-ragdoll-dirk",
          "status": "measured",
          "tokens_per_sec": 9.96
        },
        {
          "model": "Qwen3.8-27B-Uncensored-MLX",
          "made_by": "orcarouter",
          "page": "/inferences/qwen-3-8/27b~orcarouter-uncensored",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "2.8 tokens a second on Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM. About $94 per million tokens here, against OpenRouter's $1.88: the API is 50× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "Qwen 3 4B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3/4b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3-4B",
      "download_gb": 2.3,
      "params_b": 4,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 65536,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "English",
        "Tamil",
        "Telugu",
        "Hindi",
        "Malayalam"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 12,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|||"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|||",
        "machines": 1,
        "tokens_per_sec": 90.41,
        "context_tokens": null,
        "build": "IQ1_S (1-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ1_S",
        "engine": "ollama",
        "cost_usd_per_million": 2.8436,
        "cost_breakdown": {
          "tokens_per_sec": 90.41,
          "usd_per_hour": 0.9255,
          "usd_per_million": 2.8436,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "15 of 16 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 150,
              "usd_per_million": 1.6591
            },
            {
              "resource": "Memory",
              "quantity": "1.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.84,
              "usd_per_million": 0.0093
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 1.1061
            },
            {
              "resource": "Energy",
              "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 6.25,
              "usd_per_million": 0.0691
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-10T11:20:52.995Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "L7lCCEm+MI9Hu0jbJwY1iLvp9W/yHbc+9kH5QLIT9e1R8EUVc1C8ChUoEW661NrzTEZVto8F2PDvmXS+XfZjDA==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6OTAuNDA4OTM5MzYzODM1NTQsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0MzA5NjUyOTk1MDMyMTAwLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwYmM2NTBiNTlhZjg3OTQ4OTYwMzFkMmU4MmJjYTBlMTE5ZjM5MGUyMWQ2OTA5NjNhZTQ3NTc4MjM4MDg0YmY1NCIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoid2luZG93cy9hbWQ2NC9udmlkaWEvMTZnIiwibW9kZWwiOiJoZi5jby91bnNsb3RoL1F3ZW4zLTRCLUdHVUY6SVExX1MiLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEiLCJzYW1wbGVfY291bnQiOjEsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6NDIuOTI3MTcxLCJ0b3RhbF90b2tlbnMiOjM4ODF9fQ==",
          "payload_sha256": "kM8rfi2mi4PJ1JVRGUvALBPl7iHN7E1HH64q5j9ZKIQ="
        },
        "build_page": "/inferences/qwen-3/4b/unsloth-iq1_s-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3-4b-gguf",
        "weeks": [
          "2026-W35",
          "2026-W38",
          "2026-W40",
          "2026-W43"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "cheapest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": 61.33,
        "context_tokens": null,
        "build": "IQ1_S (1-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ1_S",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.2082,
        "cost_breakdown": {
          "tokens_per_sec": 61.33,
          "usd_per_hour": 0.4875,
          "usd_per_million": 2.2082,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.4892
            },
            {
              "resource": "Memory",
              "quantity": "1.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.84,
              "usd_per_million": 0.0137
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 1.6306
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.0747
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "LJXlh2taxmczw36A2OPKGSanjAHyzVHiKPpBJ+7cmTvkt0wfgJQ+R7QBExjK+qpfZT49mI6dORIf/8SkxNwlAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "q4DvIIdDBFVeIosDU+DaZPIRWL2/khEBEJqVuuY4Ym8="
        },
        "build_page": "/inferences/qwen-3/4b/unsloth-iq1_s-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3-4b-gguf",
        "weeks": [
          "2026-W35",
          "2026-W38",
          "2026-W40",
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "machine_kinds": 5,
      "machines_total": 18,
      "weeks_measured": [
        "2026-W35",
        "2026-W37",
        "2026-W38",
        "2026-W40",
        "2026-W43",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 90.41,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ1_S",
          "engine": "ollama",
          "cost_usd_per_million": 2.8436,
          "cost_breakdown": {
            "tokens_per_sec": 90.41,
            "usd_per_hour": 0.9255,
            "usd_per_million": 2.8436,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 1.6591
              },
              {
                "resource": "Memory",
                "quantity": "1.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.84,
                "usd_per_million": 0.0093
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.1061
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0691
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "L7lCCEm+MI9Hu0jbJwY1iLvp9W/yHbc+9kH5QLIT9e1R8EUVc1C8ChUoEW661NrzTEZVto8F2PDvmXS+XfZjDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "kM8rfi2mi4PJ1JVRGUvALBPl7iHN7E1HH64q5j9ZKIQ="
          },
          "build_page": "/inferences/qwen-3/4b/unsloth-iq1_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3-4b-gguf",
          "weeks": [
            "2026-W35",
            "2026-W38",
            "2026-W40",
            "2026-W43"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 80.76,
          "context_tokens": 16384,
          "build": "BF16 (16-bit, full precision) by unsloth",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:BF16",
          "engine": "ollama",
          "cost_usd_per_million": 3.2502,
          "cost_breakdown": {
            "tokens_per_sec": 80.76,
            "usd_per_hour": 0.945,
            "usd_per_million": 3.2502,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 1.8572
              },
              {
                "resource": "Memory",
                "quantity": "7.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 6.248,
                "usd_per_million": 0.0774
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.2382
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0774
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "Uw667P0enOb1mXsOG2PG/D1nlbskOfpiphmG8pZEDi22HTTrGW+/YGgcrlNVD7Xip4fE3LSpslMVVoXLiL4DCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "eaxgpNqS1poyhyGlh9rXFc2ru+qyC94yTwq3BYBIHfE="
          },
          "build_page": "/inferences/qwen-3/4b/unsloth-bf16-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3-4b-gguf",
          "weeks": [
            "2026-W38"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 61.33,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ1_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.2082,
          "cost_breakdown": {
            "tokens_per_sec": 61.33,
            "usd_per_hour": 0.4875,
            "usd_per_million": 2.2082,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.4892
              },
              {
                "resource": "Memory",
                "quantity": "1.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.84,
                "usd_per_million": 0.0137
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.6306
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0747
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "LJXlh2taxmczw36A2OPKGSanjAHyzVHiKPpBJ+7cmTvkt0wfgJQ+R7QBExjK+qpfZT49mI6dORIf/8SkxNwlAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "q4DvIIdDBFVeIosDU+DaZPIRWL2/khEBEJqVuuY4Ym8="
          },
          "build_page": "/inferences/qwen-3/4b/unsloth-iq1_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3-4b-gguf",
          "weeks": [
            "2026-W35",
            "2026-W38",
            "2026-W40",
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 53.3,
          "context_tokens": 4096,
          "build": "IQ3_XXS (3-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ3_XXS",
          "engine": "ollama",
          "cost_usd_per_million": 4.8319,
          "cost_breakdown": {
            "tokens_per_sec": 53.3,
            "usd_per_hour": 0.9272,
            "usd_per_million": 4.8319,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 2.8142
              },
              {
                "resource": "Memory",
                "quantity": "1.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.299,
                "usd_per_million": 0.0244
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.8761
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.1173
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "fqpSLr0DIktLWUNRbhlVmB6jo7kvBLgjwB4c4g0enMP59zsysO0/+5uxNCO9TnjCgP4ZRNqD6mQPOYVrOkHwDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "1HzlZGbLI9BFmVGxRIFsTOaPNX72b7Ex1ImkgOlRrg0="
          },
          "build_page": "/inferences/qwen-3/4b/unsloth-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3-4b-gguf",
          "weeks": [
            "2026-W35",
            "2026-W37",
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 50.44,
          "context_tokens": 16384,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.7067,
          "cost_breakdown": {
            "tokens_per_sec": 50.44,
            "usd_per_hour": 0.4915,
            "usd_per_million": 2.7067,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.5948
              },
              {
                "resource": "Memory",
                "quantity": "2.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.938,
                "usd_per_million": 0.0384
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.9826
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0909
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "nGqb3S0xTPGZ2AlIS2PLvtGBNbRGXG6AtA8OC+j9iOcITry7kctHeURIc1RMJ/zVIgWUphw6qxn2WY1FUMgXBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "ugBUxSE+6Hfp9Uwol9/XXPSQZdbcNuF4qSUzhRbBihI="
          },
          "build_page": "/inferences/qwen-3/4b/qwen-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-4b-gguf",
          "weeks": [
            "2026-W37",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 49.84,
          "context_tokens": 40960,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 4.3667,
          "cost_breakdown": {
            "tokens_per_sec": 49.84,
            "usd_per_hour": 0.7835,
            "usd_per_million": 4.3667,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 2.2071
              },
              {
                "resource": "Memory",
                "quantity": "2.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.938,
                "usd_per_million": 0.0389
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.0065
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.1142
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "1mCeWIXdawe2YkKNQXshmJtrsyuAk6FvaAaHkF+VVwWSE4GNfxaWHqkgqVVPs5ugisprL/wL0Ck3m7+z6kJkCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "ZS/WDMZMiN5S7xxwZSpynAE4YnfalyxMBIfEHwRJaRQ="
          },
          "build_page": "/inferences/qwen-3/4b/qwen-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-4b-gguf",
          "weeks": [
            "2026-W37",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 47.96,
          "context_tokens": 32768,
          "build": "Q5_0 (5-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.8516,
          "cost_breakdown": {
            "tokens_per_sec": 47.96,
            "usd_per_hour": 0.4924,
            "usd_per_million": 2.8516,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.6255
              },
              {
                "resource": "Memory",
                "quantity": "2.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 2.191,
                "usd_per_million": 0.0457
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.0849
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0956
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "IiTdVcTiZlLFKD9QzCavw53zoVjVXtP3Ns3ULmIviVOkF4YbuoHQxZ4DXMumJD88JQaaVxAGv5LRfNSBt20pAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "DpT4y+Od9MtoPjXdD6G66NWoZUGTydRKE0X0f/fhVB8="
          },
          "build_page": "/inferences/qwen-3/4b/qwen-q5_0-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-4b-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 47.81,
          "context_tokens": 32768,
          "build": "IQ3_XXS (3-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.8422,
          "cost_breakdown": {
            "tokens_per_sec": 47.81,
            "usd_per_hour": 0.4892,
            "usd_per_million": 2.8422,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.6275
              },
              {
                "resource": "Memory",
                "quantity": "1.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.299,
                "usd_per_million": 0.0272
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.0917
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0959
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "/pYo1R9QvwpOjh/soU8j/lqUNBXopn/VFaae7rMbtKmxqGVglsfJuqG2P7c0sc1rpBvsufyAYVM/K0RYR08UAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "0s1r4cFJ5N6j8La8jh4/kXaBuD2TdzZ5doA7Yt5Caug="
          },
          "build_page": "/inferences/qwen-3/4b/unsloth-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3-4b-gguf",
          "weeks": [
            "2026-W35",
            "2026-W37",
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 46.99,
          "context_tokens": 32768,
          "build": "Q5_K_M (5-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.9117,
          "cost_breakdown": {
            "tokens_per_sec": 46.99,
            "usd_per_hour": 0.4926,
            "usd_per_million": 2.9117,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.6384
              },
              {
                "resource": "Memory",
                "quantity": "2.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 2.242,
                "usd_per_million": 0.0477
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.128
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0975
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "rKSwONDCTRIrdAeMayWtGpPHP1LWWP2YGoGQcxPs7yic19JR7lEwcP3HiMdPKi4cdRLfizWG4p6QLT5M5JmKAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "QzCVTZCh9KyTySO6zx/xHv7hSuzkBgeDnWb74PwWaT0="
          },
          "build_page": "/inferences/qwen-3/4b/qwen-q5_k_m-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-4b-gguf",
          "weeks": [
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 38.86,
          "context_tokens": 32768,
          "build": "Q8_0 (8-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 3.5489,
          "cost_breakdown": {
            "tokens_per_sec": 38.86,
            "usd_per_hour": 0.4965,
            "usd_per_million": 3.5489,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.772
              },
              {
                "resource": "Memory",
                "quantity": "4.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.322,
                "usd_per_million": 0.0855
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.5734
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.1179
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "0mEJwHdK0Zw44csZYSklcgcS3ukop68JRN/ywbFIOXqupfLnFSW9ybbdNb+RAYNjgTnO8AboymvMQNNKM7YQBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "MfBBnhoVHnJ7vuytKpxd+qVXBJXDiDBe/CA2bvHwZsA="
          },
          "build_page": "/inferences/qwen-3/4b/qwen-q8_0-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-4b-gguf",
          "weeks": [
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 35.37,
          "context_tokens": 32768,
          "build": "Q5_K_M (5-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 7.3075,
          "cost_breakdown": {
            "tokens_per_sec": 35.37,
            "usd_per_hour": 0.9306,
            "usd_per_million": 7.3075,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 4.2405
              },
              {
                "resource": "Memory",
                "quantity": "2.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 2.242,
                "usd_per_million": 0.0634
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.827
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.1767
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "lgGV/xIplFUolwGepCLHVGUlT+0S5Hv+r4WK1+51tz9WWJB5kf1osV+9iM0LsGKMHfV84/zhD+rKzSWoMsgpBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "YUrgQW/fZIKjq9JaNS8Wm++Mnb6Kh5p13il9R2Zco3c="
          },
          "build_page": "/inferences/qwen-3/4b/qwen-q5_k_m-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-4b-gguf",
          "weeks": [
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 22.66,
          "context_tokens": 4096,
          "build": "Q5_0 (5-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 8.7188,
          "cost_breakdown": {
            "tokens_per_sec": 22.66,
            "usd_per_hour": 0.7114,
            "usd_per_million": 8.7188,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 3.9709
              },
              {
                "resource": "Memory",
                "quantity": "2.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 2.191,
                "usd_per_million": 0.0967
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 4.4121
              },
              {
                "resource": "Energy",
                "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.417,
                "usd_per_million": 0.239
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "wwU47N3va8t3ehLThYrCmHc2I3QnqTv6T6SYKU9eayS4I7DoN0x81QhpqLcsWzxQFb8z4SBw/6R0UyQ2meApAg==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MjIuNjY0NzA3ODY0NjQzMjEsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzNjE5MTEwNDQyMTA5MDAwLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwYWQyYmExZmFiYmM4ZTVmMjM4OTJjMDEzNDIyNTc5MGRkMDM1ZDY1MmNlNTQxZGMxNmUwYjA5OGM1NmFmYmI4YSIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoiZGFyd2luL2FybTY0L2FwcGxlLzE2ZyIsIm1vZGVsIjoiaGYuY28vUXdlbi9Rd2VuMy00Qi1HR1VGOlE1XzAiLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6NTEsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6NzM4Ljk0NjIxMTAwMDAwMDEsInRvdGFsX3Rva2VucyI6MTY3NDh9fQ==",
            "payload_sha256": "pDfPFm0SPAKBdK+FiIg5yz/KckIG/NEZIFB0815FBNA="
          },
          "build_page": "/inferences/qwen-3/4b/qwen-q5_0-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-4b-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W43"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 17.06,
          "context_tokens": 32768,
          "build": "Q8_0 (8-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q8_0",
          "engine": "ollama",
          "cost_usd_per_million": 15.2173,
          "cost_breakdown": {
            "tokens_per_sec": 17.06,
            "usd_per_hour": 0.9345,
            "usd_per_million": 15.2173,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 8.7937
              },
              {
                "resource": "Memory",
                "quantity": "4.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.322,
                "usd_per_million": 0.1948
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 5.8625
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.3664
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "7aM93YA3I1eAMZcXrta9FpVV8P1YxiPE7vX1eoVlPIu9cv7G7DBuZ8q/7O8pa19lN3YcipEgOage8Q+XIOQ1Aw==",
            "signing_payload_b64": "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",
            "payload_sha256": "IPNDr5TRCITlgsQvUlEUsG4vx9i0s0flLE5H0PzBFjg="
          },
          "build_page": "/inferences/qwen-3/4b/qwen-q8_0-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-4b-gguf",
          "weeks": [
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 14.37,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ1_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 13.6614,
          "cost_breakdown": {
            "tokens_per_sec": 14.37,
            "usd_per_hour": 0.7065,
            "usd_per_million": 13.6614,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 6.2649
              },
              {
                "resource": "Memory",
                "quantity": "1.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.84,
                "usd_per_million": 0.0585
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 6.961
              },
              {
                "resource": "Energy",
                "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.417,
                "usd_per_million": 0.3771
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "cFgrnzNiTJPDkGcPM0HyrdmSXep0U9UhhFrXfIIj2rM6kCy9u0wvY6w65ME8u78CYZ2KvjslVzZty2HCjHwKBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "f/sMo6oHVgWyeGG0dTH173Ajio62YDIwqV7rfiJjTjY="
          },
          "build_page": "/inferences/qwen-3/4b/unsloth-iq1_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3-4b-gguf",
          "weeks": [
            "2026-W35",
            "2026-W38",
            "2026-W40",
            "2026-W43"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 11.16,
          "context_tokens": 32768,
          "build": "Q5_0 (5-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_0",
          "engine": "ollama",
          "cost_usd_per_million": 23.165,
          "cost_breakdown": {
            "tokens_per_sec": 11.16,
            "usd_per_hour": 0.9304,
            "usd_per_million": 23.165,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 13.445
              },
              {
                "resource": "Memory",
                "quantity": "2.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 2.191,
                "usd_per_million": 0.1964
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 8.9633
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.5602
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "gKGVC0w0zcnjCGiEOMVNFVOuXbAKFoTznWkIPByFIKhnuxcRUORqTnAXvtVJCsk2K/+S7nV8YQNeV7ph4A5cAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "q6B2aQ0b7pYORXtDd1wLJBK/ROa/BnF4xMI7b2CJk0M="
          },
          "build_page": "/inferences/qwen-3/4b/qwen-q5_0-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-4b-gguf",
          "weeks": [
            "2026-W38",
            "2026-W40",
            "2026-W43"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": 4.63,
          "context_tokens": 32768,
          "build": "IQ3_XXS (3-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ3_XXS",
          "engine": "ollama",
          "cost_usd_per_million": 15.5931,
          "cost_breakdown": {
            "tokens_per_sec": 4.63,
            "usd_per_hour": 0.2602,
            "usd_per_million": 15.5931,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 15.1031
              },
              {
                "resource": "Memory",
                "quantity": "1.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.299,
                "usd_per_million": 0.2803
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.2097
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "Y59PBxXEnDHJacJxLxT+HCiEDoe34nJEOv9xMt6Y6hZV36f35FyE5ZbfZvbL3vdcW91Mq9p8u4xqxvlHJlmeAA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NC42MzQ4MTkxNDg5NTI1MTcsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0NzI1MTcyMjg5MzMyMDAwLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwM2ZlNjljZTQ0ZDdiMTQxOGRkNDI2MzMxYTUzOWYxMmMxZGU0MjhhMjM1MDE3NDQyZDQxY2VhYjBhMjllM2M3OSIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoid2luZG93cy9hbWQ2NC9pbnRlbC8xNmciLCJtb2RlbCI6ImhmLmNvL3Vuc2xvdGgvUXdlbjMtNEItR0dVRjpJUTNfWFhTIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hIiwic2FtcGxlX2NvdW50IjoyODAsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6MjI1ODYuNDI2MDU3OTk2MDA2LCJ0b3RhbF90b2tlbnMiOjEwNDY4NH19",
            "payload_sha256": "WbrJ97rZca2bL/pmCLZbmIP7nFXQ8I9bar4brJJnAPA="
          },
          "build_page": "/inferences/qwen-3/4b/unsloth-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3-4b-gguf",
          "weeks": [
            "2026-W35",
            "2026-W37",
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 65536,
          "build": "by qwen",
          "ref": "Qwen/Qwen3-4B",
          "engine": "vllm",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-20T10:09:23.154Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-3/4b/qwen-bf16-vllm",
          "hub_url": "https://huggingface.co/qwen/qwen3-4b",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-11T20:03:51.058Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-3/4b/qwen-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-4b-gguf",
          "weeks": [
            "2026-W37",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "84bf54"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3-4B"
      },
      "builds": 8,
      "variants": [],
      "bottom_line": "90 tokens a second on Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "gpt-oss 20B",
      "made_by": "openai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68783facef79a05727260de3/UPX5RQxiPGA-ZbBmArIKq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/gpt-oss/20b",
      "hub_url": "https://huggingface.co/openai/gpt-oss-20b",
      "download_gb": 10.8,
      "params_b": 20,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Assamese",
        "Bengali",
        "English",
        "Spanish",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Nepali",
        "Dutch",
        "Punjabi",
        "Russian",
        "Tamil",
        "Telugu",
        "Urdu",
        "Odia",
        "Swahili"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "Tesla T4",
          "gpu_count": 2,
          "vram_gb": 15,
          "host": "Kaggle",
          "vendor_org": "nvidia",
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
        "machines": 1,
        "tokens_per_sec": 54.73,
        "context_tokens": 65536,
        "build": "Q3_K_S (3-bit) by unsloth",
        "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q3_K_S",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.6214,
        "cost_breakdown": {
          "tokens_per_sec": 54.73,
          "usd_per_hour": 0.5165,
          "usd_per_million": 2.6214,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.5481
            },
            {
              "resource": "Memory",
              "quantity": "10.7 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 8.896,
              "usd_per_million": 0.1625
            },
            {
              "resource": "GPU",
              "quantity": "1 of 2 cards (the one this build fits on)",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 1.827
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.0837
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "6A+nYRoD6Sl7qIEUtL1lnWssfU8Lz08wJFCbGlNgM9wOLYTlTe1yQsF0KEsHk3CGmpdKxV9kkUTQWr/4uhJKBA==",
          "signing_payload_b64": "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",
          "payload_sha256": "VVj3Xe2drzC4Fx6sjqk1xAQTL6t2FAWwoyOz2QTxMok="
        },
        "build_page": "/inferences/gpt-oss/20b/unsloth-q3_k_s-ollama",
        "hub_url": "https://huggingface.co/unsloth/gpt-oss-20b-gguf",
        "weeks": [
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": {
        "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "Tesla T4",
          "gpu_count": 2,
          "vram_gb": 15,
          "host": "Kaggle",
          "vendor_org": "nvidia",
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
        "machines": 1,
        "tokens_per_sec": 54.73,
        "context_tokens": 65536,
        "build": "Q3_K_S (3-bit) by unsloth",
        "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q3_K_S",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.6214,
        "cost_breakdown": {
          "tokens_per_sec": 54.73,
          "usd_per_hour": 0.5165,
          "usd_per_million": 2.6214,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.5481
            },
            {
              "resource": "Memory",
              "quantity": "10.7 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 8.896,
              "usd_per_million": 0.1625
            },
            {
              "resource": "GPU",
              "quantity": "1 of 2 cards (the one this build fits on)",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 1.827
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.0837
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "6A+nYRoD6Sl7qIEUtL1lnWssfU8Lz08wJFCbGlNgM9wOLYTlTe1yQsF0KEsHk3CGmpdKxV9kkUTQWr/4uhJKBA==",
          "signing_payload_b64": "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",
          "payload_sha256": "VVj3Xe2drzC4Fx6sjqk1xAQTL6t2FAWwoyOz2QTxMok="
        },
        "build_page": "/inferences/gpt-oss/20b/unsloth-q3_k_s-ollama",
        "hub_url": "https://huggingface.co/unsloth/gpt-oss-20b-gguf",
        "weeks": [
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "machine_kinds": 3,
      "machines_total": 17,
      "weeks_measured": [
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "Tesla T4",
            "gpu_count": 2,
            "vram_gb": 15,
            "host": "Kaggle",
            "vendor_org": "nvidia",
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
          "machines": 1,
          "tokens_per_sec": 54.73,
          "context_tokens": 65536,
          "build": "Q3_K_S (3-bit) by unsloth",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.6214,
          "cost_breakdown": {
            "tokens_per_sec": 54.73,
            "usd_per_hour": 0.5165,
            "usd_per_million": 2.6214,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.5481
              },
              {
                "resource": "Memory",
                "quantity": "10.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 8.896,
                "usd_per_million": 0.1625
              },
              {
                "resource": "GPU",
                "quantity": "1 of 2 cards (the one this build fits on)",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.827
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0837
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "6A+nYRoD6Sl7qIEUtL1lnWssfU8Lz08wJFCbGlNgM9wOLYTlTe1yQsF0KEsHk3CGmpdKxV9kkUTQWr/4uhJKBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "VVj3Xe2drzC4Fx6sjqk1xAQTL6t2FAWwoyOz2QTxMok="
          },
          "build_page": "/inferences/gpt-oss/20b/unsloth-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/gpt-oss-20b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "Tesla T4",
            "gpu_count": 2,
            "vram_gb": 15,
            "host": "Kaggle",
            "vendor_org": "nvidia",
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
          "machines": 1,
          "tokens_per_sec": 52.39,
          "context_tokens": 65536,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.7409,
          "cost_breakdown": {
            "tokens_per_sec": 52.39,
            "usd_per_hour": 0.517,
            "usd_per_million": 2.7409,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.5726
              },
              {
                "resource": "Memory",
                "quantity": "10.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 9.022,
                "usd_per_million": 0.1722
              },
              {
                "resource": "GPU",
                "quantity": "1 of 2 cards (the one this build fits on)",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.9087
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0875
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 312.8,
              "multi_core_hashes_per_sec": 618.9,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T21:33:01.078Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 2.4,
              "latency_ns": 21.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T21:33:01.673Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 21.6,
              "watts": 59.1,
              "tflops_per_watt": 0.365,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-02T21:43:16.805Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4293.1,
              "write_mb_s": 60.4,
              "random_read_iops": 514568,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-02T21:33:02.973Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T00:21:51.818Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220d79e72fae9f23f0437b58a3411e6bdb14e5c77f568cc3bdf6622534463c76d30",
            "signature": "ZlROyZ5tTzYRW8z97CG94RXvxMBBG6cWEdV5YWlAoXc9CjyiqOhhhaTpTC0n446Toubz3OXNrwbYfS9arj9GAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "jEwoDxq9KJMQ9zYLF0Zr40bMTfI90tVTbe4Z8vZUN9E="
          },
          "build_page": "/inferences/gpt-oss/20b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gpt-oss-20b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "c76d30"
          ]
        },
        {
          "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "Tesla T4",
            "gpu_count": 2,
            "vram_gb": 15,
            "host": "Kaggle",
            "vendor_org": "nvidia",
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
          "machines": 1,
          "tokens_per_sec": 44.16,
          "context_tokens": 131072,
          "build": "Q8_K_XL (8-bit) by unsloth",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q8_K_XL",
          "engine": "ollama-host",
          "cost_usd_per_million": 3.2797,
          "cost_breakdown": {
            "tokens_per_sec": 44.16,
            "usd_per_hour": 0.5214,
            "usd_per_million": 3.2797,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.6794
              },
              {
                "resource": "Memory",
                "quantity": "12.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 10.241,
                "usd_per_million": 0.2319
              },
              {
                "resource": "GPU",
                "quantity": "1 of 2 cards (the one this build fits on)",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.2646
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.1038
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "oiUJV6I1RqoOEmrMo8b90V6Ri8LJeIqyOWgEz3ICAqJ3NBv9enH/SP1I0ZL9I+fBqFA1eb45GReehF+bldvECA==",
            "signing_payload_b64": "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",
            "payload_sha256": "jlOYTMHrjhM6BwDbbBd0wJWc/UgODgcIwx4JwdNq2js="
          },
          "build_page": "/inferences/gpt-oss/20b/unsloth-q8_k_xl-ollama",
          "hub_url": "https://huggingface.co/unsloth/gpt-oss-20b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 24.03,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 9.3514,
          "cost_breakdown": {
            "tokens_per_sec": 24.03,
            "usd_per_hour": 0.809,
            "usd_per_million": 9.3514,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 4.5776
              },
              {
                "resource": "Memory",
                "quantity": "10.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 9.022,
                "usd_per_million": 0.3754
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 4.1614
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.237
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "rzy7bhjq6GLr+VmK5W2qAfcs7X9nEp4dC6HMepoA2QgEdvS47oX1SkP/92P+Ca4bjy09aCwmBG+cxDKIcyWhCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "fV8qxbnMEuLlnWiKrTVxCNICsRJi1opdVXqscaeX924="
          },
          "build_page": "/inferences/gpt-oss/20b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gpt-oss-20b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 3,
          "tokens_per_sec": 4.56,
          "context_tokens": 4096,
          "build": "Q4_0 (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 8.6295,
          "cost_breakdown": {
            "tokens_per_sec": 4.56,
            "usd_per_hour": 0.1416,
            "usd_per_million": 8.6295,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.5802
              },
              {
                "resource": "Memory",
                "quantity": "10.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 8.926,
                "usd_per_million": 1.9578
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0915
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 185.9,
              "multi_core_hashes_per_sec": 588,
              "sustained_ratio": 1.4,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-14T19:37:12.663Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 1.7,
              "latency_ns": 71.3,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-14T19:37:15.832Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2202.4,
              "write_mb_s": 15.5,
              "random_read_iops": 196685,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-14T19:37:24.078Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T20:22:46.948Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "YyzbZTOjhiM+NIhCD3X67fna6xFNzI6VTHGQZmRY5zqb6y1sSFwAXFlW05zBBwC2U2cAsQy6pb3KGri4hlWVCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "CfyUetf0TNjuJqVUDuD7MgL2vm6jU9oWa4MsYSUBfv4="
          },
          "build_page": "/inferences/gpt-oss/20b/unsloth-q4_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gpt-oss-20b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "4e94b2",
            "662a4e",
            "e3654f"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 3,
          "tokens_per_sec": 4.41,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 8.9491,
          "cost_breakdown": {
            "tokens_per_sec": 4.41,
            "usd_per_hour": 0.142,
            "usd_per_million": 8.9491,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.8073
              },
              {
                "resource": "Memory",
                "quantity": "10.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 9.022,
                "usd_per_million": 2.0472
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0946
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T06:35:24.096Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "+G+ybq6tBWwm8EeOJZCgzn79CShJvim8ErkVPpzvObHOUhDPdB/KA3o9E7uGP6DiI32t9OYZET9XE/9EhybnCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "0R0oWGjp4U8rA980GTXB+sz8mG6Wr/4Ig/X4ubJEbIc="
          },
          "build_page": "/inferences/gpt-oss/20b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gpt-oss-20b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "e3654f",
            "e61d3b",
            "662a4e"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 2,
          "tokens_per_sec": 4.35,
          "context_tokens": 4096,
          "build": "Q3_K_S (3-bit) by unsloth",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 9.032,
          "cost_breakdown": {
            "tokens_per_sec": 4.35,
            "usd_per_hour": 0.1415,
            "usd_per_million": 9.032,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.8924
              },
              {
                "resource": "Memory",
                "quantity": "10.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 8.896,
                "usd_per_million": 2.0438
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0958
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:44:44.996Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "F0s6Gh5rDuyr4vEDOeR1kqpH1W+yQ1M/KH/J6iKXQs29UGm/mSrfN47JCnAyS5hxolalU0U9RZH9rjzTd/GyBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "x/e4uK04T8R7jWxRm8p5HcufuBBn/lOtxOOeyjHRSI4="
          },
          "build_page": "/inferences/gpt-oss/20b/unsloth-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/gpt-oss-20b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e",
            "e3654f"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 2,
          "tokens_per_sec": 4.16,
          "context_tokens": 4096,
          "build": "Q2_K (2-bit) by unsloth",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q2_K",
          "engine": "ollama-host",
          "cost_usd_per_million": 9.4507,
          "cost_breakdown": {
            "tokens_per_sec": 4.16,
            "usd_per_hour": 0.1415,
            "usd_per_million": 9.4507,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 7.211
              },
              {
                "resource": "Memory",
                "quantity": "10.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 8.901,
                "usd_per_million": 2.1395
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1002
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:44:44.996Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "RKQoJp5fVa+d3jp+CBOHgEaNM9UvrXNUuyjlcWyMO+/FeAH2Fgy0mBtD7BbaEm+DjSOG8Wbuqhl62ta43+WJDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "PUqF+vuaVswGiydPi/94HmxVrP1bGe4m3jIO/DhOaNI="
          },
          "build_page": "/inferences/gpt-oss/20b/unsloth-q2_k-ollama",
          "hub_url": "https://huggingface.co/unsloth/gpt-oss-20b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e",
            "e3654f"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 2,
          "tokens_per_sec": 4.04,
          "context_tokens": 4096,
          "build": "Q8_0 (8-bit) by unsloth",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 9.856,
          "cost_breakdown": {
            "tokens_per_sec": 4.04,
            "usd_per_hour": 0.1433,
            "usd_per_million": 9.856,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 7.4264
              },
              {
                "resource": "Memory",
                "quantity": "11.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 9.398,
                "usd_per_million": 2.3264
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1032
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T06:35:24.096Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "gi+quH7r/k3h8tXvHURih+O0v8Tf50UIU4u1VlA9rESv5+aX1UyiHHUqhu7c7O0vsxemXusW0clM3VGWg9PbBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "F1WzqmDROeCS7UKDO3BgqNkwhlmvZDMBhI7bj2slhgI="
          },
          "build_page": "/inferences/gpt-oss/20b/unsloth-q8_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gpt-oss-20b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "e3654f",
            "662a4e"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 3.11,
          "context_tokens": 4096,
          "build": "Q8_K_XL (8-bit) by unsloth",
          "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q8_K_XL",
          "engine": "ollama-host",
          "cost_usd_per_million": 13.0639,
          "cost_breakdown": {
            "tokens_per_sec": 3.11,
            "usd_per_hour": 0.1464,
            "usd_per_million": 13.0639,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 9.6394
              },
              {
                "resource": "Memory",
                "quantity": "12.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 10.241,
                "usd_per_million": 3.2906
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.134
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:44:44.996Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "1GSxD6baWLSM851C1oa8ybWrh95aBbcsfdqewrzQ6TTRj3Y2Zn537hHispEQ8kt+PzCS8HxzJHjeB9EhmFe2Cw==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6My4xMTIyMzM4MTcwODUxNDczLCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5NDcyNTA4NDk5Njg5ODc1Niwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDZlZWY2YjczODRiNjY0MjgwZjdhNjIyYjZjMWE4MzFhNDUzMDI5ZDA5ODA5NTZlYzgwOTA5NTRlNzI2NjJhNGUiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L25vbmUvMzJnIiwibW9kZWwiOiJoZi5jby91bnNsb3RoL2dwdC1vc3MtMjBiLUdHVUY6UThfS19YTCIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo0NCwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjoyNTM4LjY5MTAwNTk5OSwidG90YWxfdG9rZW5zIjo3OTAxfX0=",
            "payload_sha256": "U4CHi8mB9sRxoqBbkYJ68ExeDztURvrQxXlk5nVtbe0="
          },
          "build_page": "/inferences/gpt-oss/20b/unsloth-q8_k_xl-ollama",
          "hub_url": "https://huggingface.co/unsloth/gpt-oss-20b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "openai/gpt-oss-20b",
        "url": "https://openrouter.ai/openai/gpt-oss-20b",
        "input_usd_per_million": 0.03,
        "output_usd_per_million": 0.13,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.02,
            "usd_per_m_out": 0.03,
            "context_length": 131072,
            "uptime_1d": 99.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.018,
            "usd_per_m_out": 0.09,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.018,
            "usd_per_m_out": 0.09,
            "context_length": 131072,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16/flex",
            "quantization": "bf16",
            "usd_per_m_in": 0.024,
            "usd_per_m_out": 0.112,
            "context_length": 131072,
            "uptime_1d": 91.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.13,
            "context_length": 131072,
            "uptime_1d": 93.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.029,
            "usd_per_m_out": 0.14,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.14,
            "context_length": 131072,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.15,
            "context_length": 131072,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock/eu-west-1",
            "quantization": null,
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.15,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock",
            "quantization": null,
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.15,
            "context_length": 131072,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 0.18,
            "context_length": 131072,
            "uptime_1d": 90.6,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Groq",
            "tag": "groq",
            "quantization": null,
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.3,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 12
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openai/gpt-oss-20b"
      },
      "builds": 8,
      "variants": [
        {
          "model": "gpt-oss-20b-heretic",
          "made_by": "p-e-w",
          "page": "/inferences/gpt-oss/20b~p-e-w-heretic",
          "status": "measured",
          "tokens_per_sec": 3.36
        },
        {
          "model": "Huihui-gpt-oss-20b-BF16-abliterated",
          "made_by": "huihui-ai",
          "page": "/inferences/gpt-oss/20b~huihui-ai-abliterated",
          "status": "measured",
          "tokens_per_sec": 3.28
        },
        {
          "model": "gpt-oss-20b-Derestricted",
          "made_by": "ArliAI",
          "page": "/inferences/gpt-oss/20b~arliai-derestricted",
          "status": "measured",
          "tokens_per_sec": 3.44
        },
        {
          "model": "openai_gpt-oss-20b-GGUF-MXFP4-Experimental",
          "made_by": "bartowski",
          "page": "/inferences/gpt-oss/20b~bartowski-openai-experimental",
          "status": "measured",
          "tokens_per_sec": 4.32
        },
        {
          "model": "OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
          "made_by": "DavidAU",
          "page": "/inferences/gpt-oss/20b~davidau-openai-claude-4-5-opus-heretic-uncensored",
          "status": "measured",
          "tokens_per_sec": 4.82
        }
      ],
      "bottom_line": "55 tokens a second on Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM. About $2.62 per million tokens here, against OpenRouter's $0.13: the API is 20× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "GLM-4.7 Flash",
      "made_by": "zai-org",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/glm-4-7/flash",
      "hub_url": "https://huggingface.co/zai-org/GLM-4.7-Flash",
      "download_gb": 15.7,
      "params_b": 28,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Assamese",
        "Bengali",
        "English",
        "Spanish",
        "Hindi",
        "Malayalam",
        "Marathi",
        "Nepali",
        "Dutch",
        "Odia",
        "Punjabi",
        "Russian",
        "Tamil",
        "Telugu",
        "Urdu",
        "Kannada"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "Tesla T4",
          "gpu_count": 2,
          "vram_gb": 15,
          "host": "Kaggle",
          "vendor_org": "nvidia",
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
        "machines": 1,
        "tokens_per_sec": 44.98,
        "context_tokens": 131072,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/GLM-4.7-Flash-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 5.6235,
        "cost_breakdown": {
          "tokens_per_sec": 44.98,
          "usd_per_hour": 0.9107,
          "usd_per_million": 5.6235,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.6669
            },
            {
              "resource": "Memory",
              "quantity": "17.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 14.212,
              "usd_per_million": 0.3159
            },
            {
              "resource": "GPU",
              "quantity": "2 of 2 cards (the cards this build is split across)",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 200,
              "usd_per_million": 4.4461
            },
            {
              "resource": "Energy",
              "quantity": "315 W modelled (3 threads × 5 W + 2 × 150 W cards)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 8.75,
              "usd_per_million": 0.1945
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "XUEyipid46+kLBdt3vsYkzJSj7lYh1XmPL/r4O2WnoJJpl0lOCh1PyxxpoU9dCw5dAHz1NzB/nNmVCgSc8TTDw==",
          "signing_payload_b64": "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",
          "payload_sha256": "TmLvZzzo9pfzI8v17eOhsuWijolaIu8VdUXR11005UQ="
        },
        "build_page": "/inferences/glm-4-7/flash/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/glm-4.7-flash-gguf",
        "weeks": [
          "2026-W44",
          "2026-W45"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": {
        "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "Tesla T4",
          "gpu_count": 2,
          "vram_gb": 15,
          "host": "Kaggle",
          "vendor_org": "nvidia",
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
        "machines": 1,
        "tokens_per_sec": 44.98,
        "context_tokens": 131072,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/GLM-4.7-Flash-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 5.6235,
        "cost_breakdown": {
          "tokens_per_sec": 44.98,
          "usd_per_hour": 0.9107,
          "usd_per_million": 5.6235,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.6669
            },
            {
              "resource": "Memory",
              "quantity": "17.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 14.212,
              "usd_per_million": 0.3159
            },
            {
              "resource": "GPU",
              "quantity": "2 of 2 cards (the cards this build is split across)",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 200,
              "usd_per_million": 4.4461
            },
            {
              "resource": "Energy",
              "quantity": "315 W modelled (3 threads × 5 W + 2 × 150 W cards)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 8.75,
              "usd_per_million": 0.1945
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "XUEyipid46+kLBdt3vsYkzJSj7lYh1XmPL/r4O2WnoJJpl0lOCh1PyxxpoU9dCw5dAHz1NzB/nNmVCgSc8TTDw==",
          "signing_payload_b64": "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",
          "payload_sha256": "TmLvZzzo9pfzI8v17eOhsuWijolaIu8VdUXR11005UQ="
        },
        "build_page": "/inferences/glm-4-7/flash/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/glm-4.7-flash-gguf",
        "weeks": [
          "2026-W44",
          "2026-W45"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "machine_kinds": 4,
      "machines_total": 5,
      "weeks_measured": [
        "2026-W44",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "Tesla T4",
            "gpu_count": 2,
            "vram_gb": 15,
            "host": "Kaggle",
            "vendor_org": "nvidia",
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
          "machines": 1,
          "tokens_per_sec": 44.98,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/GLM-4.7-Flash-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 5.6235,
          "cost_breakdown": {
            "tokens_per_sec": 44.98,
            "usd_per_hour": 0.9107,
            "usd_per_million": 5.6235,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.6669
              },
              {
                "resource": "Memory",
                "quantity": "17.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 14.212,
                "usd_per_million": 0.3159
              },
              {
                "resource": "GPU",
                "quantity": "2 of 2 cards (the cards this build is split across)",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 200,
                "usd_per_million": 4.4461
              },
              {
                "resource": "Energy",
                "quantity": "315 W modelled (3 threads × 5 W + 2 × 150 W cards)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 8.75,
                "usd_per_million": 0.1945
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "XUEyipid46+kLBdt3vsYkzJSj7lYh1XmPL/r4O2WnoJJpl0lOCh1PyxxpoU9dCw5dAHz1NzB/nNmVCgSc8TTDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "TmLvZzzo9pfzI8v17eOhsuWijolaIu8VdUXR11005UQ="
          },
          "build_page": "/inferences/glm-4-7/flash/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/glm-4.7-flash-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 40.58,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/GLM-4.7-Flash-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 5.6652,
          "cost_breakdown": {
            "tokens_per_sec": 40.58,
            "usd_per_hour": 0.8277,
            "usd_per_million": 5.6652,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 2.7106
              },
              {
                "resource": "Memory",
                "quantity": "17.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 14.212,
                "usd_per_million": 0.3502
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.4642
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.1403
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "qOoNAso8oxKI5fA5h/+97Jp0fx1BfG9SK1selBw9lt6PEkaRaQCIlzcFIvKaA0KjcVkBZ55MVjX1mVL/yUifAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "oVORjB2am+l7Mp5SaeLxQ+bEdxG6eFMMo1+WE0OpX/Q="
          },
          "build_page": "/inferences/glm-4-7/flash/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/glm-4.7-flash-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 2,
          "tokens_per_sec": 3.55,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/GLM-4.7-Flash-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 12.5701,
          "cost_breakdown": {
            "tokens_per_sec": 3.55,
            "usd_per_hour": 0.1607,
            "usd_per_million": 12.5701,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 8.4498
              },
              {
                "resource": "Memory",
                "quantity": "17.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 14.212,
                "usd_per_million": 4.0029
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1175
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "xGfHrry32UfdIWxUCLLWOfeG3HgDHy04wN3bxNtJbQAlnvDAGvkMyU7Ylr7eRz0kwhHxyVWo93aRzVURcEogBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "M+qXLEAZFbxdEpXxUK7hawQRcB1vxQyNmAGjYWGeEf0="
          },
          "build_page": "/inferences/glm-4-7/flash/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/glm-4.7-flash-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "e61d3b",
            "e3654f"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 3.21,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/GLM-4.7-Flash-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 13.9098,
          "cost_breakdown": {
            "tokens_per_sec": 3.21,
            "usd_per_hour": 0.1607,
            "usd_per_million": 13.9098,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 9.3503
              },
              {
                "resource": "Memory",
                "quantity": "17.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 14.212,
                "usd_per_million": 4.4296
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.13
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:44:44.996Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "osyHIjL41qBWM+G7elAn5fr8RAgFn18d2ZQgtdX08uWiS+YSvNASzqEvHGfZrdVsrV3J+6qUTz496gA71/x/Aw==",
            "signing_payload_b64": "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",
            "payload_sha256": "ZVTi0FWgwZzqp36XBQD39h1uc9u3k3rCtFFanU0NkzI="
          },
          "build_page": "/inferences/glm-4-7/flash/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/glm-4.7-flash-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-4.7-flash",
        "url": "https://openrouter.ai/z-ai/glm-4.7-flash",
        "input_usd_per_million": 0.0605,
        "output_usd_per_million": 0.4,
        "free_tier": false,
        "context_tokens": 200000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.0605,
            "usd_per_m_out": 0.4,
            "context_length": 131072,
            "uptime_1d": 99.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.4,
            "context_length": 200000,
            "uptime_1d": 5.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-4.7-Flash"
      },
      "builds": 3,
      "variants": [
        {
          "model": "GLM-4.7-Flash-RAM-20GB-MLX",
          "made_by": "baa-ai",
          "page": "/inferences/glm-4-7/flash~baa-ai-ram-20gb",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "Huihui-GLM-4.7-Flash-abliterated-mlx-4bit",
          "made_by": "huihui-ai",
          "page": "/inferences/glm-4-7/flash~huihui-ai-abliterated",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "45 tokens a second on Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM. About $5.62 per million tokens here, against OpenRouter's $0.40: the API is 14× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "Qwen 3.5 4B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3-5/4b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.5-4B",
      "download_gb": 2.6,
      "params_b": 4,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 262144,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 41.82,
        "context_tokens": 262144,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3.5-4B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 5.209,
        "cost_breakdown": {
          "tokens_per_sec": 41.82,
          "usd_per_hour": 0.7842,
          "usd_per_million": 5.209,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 2.6305
            },
            {
              "resource": "Memory",
              "quantity": "2.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 2.126,
              "usd_per_million": 0.0508
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 2.3914
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.1362
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "93zGjNlrQWFWFFfbgbTtz8mrqrntDDMrC+k7Aw+iSwdFVBuCZm1PUnuDezqXz9ogdPJLvOoe2EyPCBY7KYsYDg==",
          "signing_payload_b64": "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",
          "payload_sha256": "GtCHeNIu4MOolLC6eTFXMdSUR9f3BcyoMcaR+VeO4mA="
        },
        "build_page": "/inferences/qwen-3-5/4b/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3.5-4b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 41.82,
        "context_tokens": 262144,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3.5-4B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 5.209,
        "cost_breakdown": {
          "tokens_per_sec": 41.82,
          "usd_per_hour": 0.7842,
          "usd_per_million": 5.209,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 2.6305
            },
            {
              "resource": "Memory",
              "quantity": "2.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 2.126,
              "usd_per_million": 0.0508
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 2.3914
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.1362
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "93zGjNlrQWFWFFfbgbTtz8mrqrntDDMrC+k7Aw+iSwdFVBuCZm1PUnuDezqXz9ogdPJLvOoe2EyPCBY7KYsYDg==",
          "signing_payload_b64": "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",
          "payload_sha256": "GtCHeNIu4MOolLC6eTFXMdSUR9f3BcyoMcaR+VeO4mA="
        },
        "build_page": "/inferences/qwen-3-5/4b/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3.5-4b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 41.82,
          "context_tokens": 262144,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3.5-4B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 5.209,
          "cost_breakdown": {
            "tokens_per_sec": 41.82,
            "usd_per_hour": 0.7842,
            "usd_per_million": 5.209,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 2.6305
              },
              {
                "resource": "Memory",
                "quantity": "2.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 2.126,
                "usd_per_million": 0.0508
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.3914
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.1362
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "93zGjNlrQWFWFFfbgbTtz8mrqrntDDMrC+k7Aw+iSwdFVBuCZm1PUnuDezqXz9ogdPJLvOoe2EyPCBY7KYsYDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "GtCHeNIu4MOolLC6eTFXMdSUR9f3BcyoMcaR+VeO4mA="
          },
          "build_page": "/inferences/qwen-3-5/4b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.5-4b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.5-4B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "42 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Qianfan-OCR Base",
      "made_by": "baidu",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/64f187a2cc1c03340ac30498/TYYUxK8xD1AxExFMWqbZD.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qianfan-ocr/base",
      "hub_url": "https://huggingface.co/baidu/Qianfan-OCR",
      "download_gb": 2.5,
      "params_b": 4.5,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": 32768,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "English",
        "Hindi",
        "Kannada",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 17.56,
        "context_tokens": 32768,
        "build": "Q4_K_M (4-bit) by Reza2kn",
        "ref": "hf.co/Reza2kn/Qianfan-OCR-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 12.4008,
        "cost_breakdown": {
          "tokens_per_sec": 17.56,
          "usd_per_hour": 0.7841,
          "usd_per_million": 12.4008,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 6.2628
            },
            {
              "resource": "Memory",
              "quantity": "2.5 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 2.113,
              "usd_per_million": 0.1203
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 5.6935
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.3242
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "94bsoxIscQhgm/aSN+mpOeTO8FAXCcTcZos1xpksWmyNS7JSJQAUaURyZxxrExmvwKhvasBJ7Zb1IZYkqhELCg==",
          "signing_payload_b64": "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",
          "payload_sha256": "ZljfoSOoxrrfF2AtTC3UHMPyaKjFZXsv6RPkq1WLO5k="
        },
        "build_page": "/inferences/qianfan-ocr/base/reza2kn-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/reza2kn/qianfan-ocr-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 17.56,
        "context_tokens": 32768,
        "build": "Q4_K_M (4-bit) by Reza2kn",
        "ref": "hf.co/Reza2kn/Qianfan-OCR-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 12.4008,
        "cost_breakdown": {
          "tokens_per_sec": 17.56,
          "usd_per_hour": 0.7841,
          "usd_per_million": 12.4008,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 6.2628
            },
            {
              "resource": "Memory",
              "quantity": "2.5 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 2.113,
              "usd_per_million": 0.1203
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 5.6935
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.3242
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "94bsoxIscQhgm/aSN+mpOeTO8FAXCcTcZos1xpksWmyNS7JSJQAUaURyZxxrExmvwKhvasBJ7Zb1IZYkqhELCg==",
          "signing_payload_b64": "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",
          "payload_sha256": "ZljfoSOoxrrfF2AtTC3UHMPyaKjFZXsv6RPkq1WLO5k="
        },
        "build_page": "/inferences/qianfan-ocr/base/reza2kn-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/reza2kn/qianfan-ocr-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 17.56,
          "context_tokens": 32768,
          "build": "Q4_K_M (4-bit) by Reza2kn",
          "ref": "hf.co/Reza2kn/Qianfan-OCR-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 12.4008,
          "cost_breakdown": {
            "tokens_per_sec": 17.56,
            "usd_per_hour": 0.7841,
            "usd_per_million": 12.4008,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 6.2628
              },
              {
                "resource": "Memory",
                "quantity": "2.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 2.113,
                "usd_per_million": 0.1203
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 5.6935
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.3242
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "94bsoxIscQhgm/aSN+mpOeTO8FAXCcTcZos1xpksWmyNS7JSJQAUaURyZxxrExmvwKhvasBJ7Zb1IZYkqhELCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "ZljfoSOoxrrfF2AtTC3UHMPyaKjFZXsv6RPkq1WLO5k="
          },
          "build_page": "/inferences/qianfan-ocr/base/reza2kn-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/reza2kn/qianfan-ocr-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/baidu/Qianfan-OCR"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "18 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Granite 4.2 3B",
      "made_by": "ibm-granite",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/639bcaa2445b133a4e942436/CEW-OjXkRkDNmTxSu8Egh.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/granite-4-2/3b",
      "hub_url": "https://huggingface.co/ibm-granite/granite-4.2-3b",
      "download_gb": 2.1,
      "params_b": 3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 32768,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "English",
        "Kannada",
        "Tamil"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 16.87,
        "context_tokens": 32768,
        "build": "Q4_K_M (4-bit) by ibm-granite",
        "ref": "hf.co/ibm-granite/granite-4.2-3b-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 12.8887,
        "cost_breakdown": {
          "tokens_per_sec": 16.87,
          "usd_per_hour": 0.7828,
          "usd_per_million": 12.8887,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 6.5203
            },
            {
              "resource": "Memory",
              "quantity": "2.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.742,
              "usd_per_million": 0.1033
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 5.9276
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.3375
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "GTgUE+M0YNBrhCd+nfhxvfuUTHXQltxaRxxqr+BRb3X6swidzvpDLWqIJLj49oUitPT6RumT6tXbUpxWniVtAQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "g8K0PFW88Jhe5Bfg9xesFajjxiyXUWNxUwNdxY7RDHc="
        },
        "build_page": "/inferences/granite-4-2/3b/ibm-granite-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/ibm-granite/granite-4.2-3b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 16.87,
        "context_tokens": 32768,
        "build": "Q4_K_M (4-bit) by ibm-granite",
        "ref": "hf.co/ibm-granite/granite-4.2-3b-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 12.8887,
        "cost_breakdown": {
          "tokens_per_sec": 16.87,
          "usd_per_hour": 0.7828,
          "usd_per_million": 12.8887,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 6.5203
            },
            {
              "resource": "Memory",
              "quantity": "2.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.742,
              "usd_per_million": 0.1033
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 5.9276
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.3375
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "GTgUE+M0YNBrhCd+nfhxvfuUTHXQltxaRxxqr+BRb3X6swidzvpDLWqIJLj49oUitPT6RumT6tXbUpxWniVtAQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "g8K0PFW88Jhe5Bfg9xesFajjxiyXUWNxUwNdxY7RDHc="
        },
        "build_page": "/inferences/granite-4-2/3b/ibm-granite-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/ibm-granite/granite-4.2-3b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 16.87,
          "context_tokens": 32768,
          "build": "Q4_K_M (4-bit) by ibm-granite",
          "ref": "hf.co/ibm-granite/granite-4.2-3b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 12.8887,
          "cost_breakdown": {
            "tokens_per_sec": 16.87,
            "usd_per_hour": 0.7828,
            "usd_per_million": 12.8887,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 6.5203
              },
              {
                "resource": "Memory",
                "quantity": "2.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.742,
                "usd_per_million": 0.1033
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 5.9276
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.3375
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "GTgUE+M0YNBrhCd+nfhxvfuUTHXQltxaRxxqr+BRb3X6swidzvpDLWqIJLj49oUitPT6RumT6tXbUpxWniVtAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "g8K0PFW88Jhe5Bfg9xesFajjxiyXUWNxUwNdxY7RDHc="
          },
          "build_page": "/inferences/granite-4-2/3b/ibm-granite-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/ibm-granite/granite-4.2-3b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/ibm-granite/granite-4.2-3b"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "17 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Qwen 3 0.6B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3/0.6b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3-0.6B",
      "download_gb": 0.6,
      "params_b": 0.6,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 12.79,
        "context_tokens": 4096,
        "build": "Q8_0 (8-bit) by Qwen",
        "ref": "hf.co/Qwen/Qwen3-0.6B-GGUF:Q8_0",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.4165,
        "cost_breakdown": {
          "tokens_per_sec": 12.79,
          "usd_per_hour": 0.1113,
          "usd_per_million": 2.4165,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 2.3452
            },
            {
              "resource": "Memory",
              "quantity": "0.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.496,
              "usd_per_million": 0.0388
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0326
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "niteGoX8PYMVwf9p1TUNZ+ozgvTdbaqpwKJZjOxEaXGx6nRadPEwy6sReyDKL0kLk4kOSo57EH7cWqQCIwS6DA==",
          "signing_payload_b64": "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",
          "payload_sha256": "yv+sghfDR8TIwnGaA2+/5F4A0PTXQFHb7SWc3NQdQ6k="
        },
        "build_page": "/inferences/qwen-3/0.6b/qwen-q8_0-ollama",
        "hub_url": "https://huggingface.co/qwen/qwen3-0.6b-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 12.79,
        "context_tokens": 4096,
        "build": "Q8_0 (8-bit) by Qwen",
        "ref": "hf.co/Qwen/Qwen3-0.6B-GGUF:Q8_0",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.4165,
        "cost_breakdown": {
          "tokens_per_sec": 12.79,
          "usd_per_hour": 0.1113,
          "usd_per_million": 2.4165,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 2.3452
            },
            {
              "resource": "Memory",
              "quantity": "0.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.496,
              "usd_per_million": 0.0388
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0326
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "niteGoX8PYMVwf9p1TUNZ+ozgvTdbaqpwKJZjOxEaXGx6nRadPEwy6sReyDKL0kLk4kOSo57EH7cWqQCIwS6DA==",
          "signing_payload_b64": "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",
          "payload_sha256": "yv+sghfDR8TIwnGaA2+/5F4A0PTXQFHb7SWc3NQdQ6k="
        },
        "build_page": "/inferences/qwen-3/0.6b/qwen-q8_0-ollama",
        "hub_url": "https://huggingface.co/qwen/qwen3-0.6b-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W34",
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 12.79,
          "context_tokens": 4096,
          "build": "Q8_0 (8-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-0.6B-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.4165,
          "cost_breakdown": {
            "tokens_per_sec": 12.79,
            "usd_per_hour": 0.1113,
            "usd_per_million": 2.4165,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.3452
              },
              {
                "resource": "Memory",
                "quantity": "0.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.496,
                "usd_per_million": 0.0388
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0326
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "niteGoX8PYMVwf9p1TUNZ+ozgvTdbaqpwKJZjOxEaXGx6nRadPEwy6sReyDKL0kLk4kOSo57EH7cWqQCIwS6DA==",
            "signing_payload_b64": "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",
            "payload_sha256": "yv+sghfDR8TIwnGaA2+/5F4A0PTXQFHb7SWc3NQdQ6k="
          },
          "build_page": "/inferences/qwen-3/0.6b/qwen-q8_0-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-0.6b-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3-0.6B"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "13 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Qwen 3 VL-4B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3/vl-4b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3-VL-4B-Instruct",
      "download_gb": 7.5,
      "params_b": 4,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "English",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Tamil",
        "Telugu",
        "Dutch",
        "Russian",
        "Punjabi"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": 11.91,
        "context_tokens": 131072,
        "build": "F16 (16-bit, full precision) by Qwen",
        "ref": "hf.co/Qwen/Qwen3-VL-4B-Instruct-GGUF:F16",
        "engine": "ollama-host",
        "cost_usd_per_million": 11.8296,
        "cost_breakdown": {
          "tokens_per_sec": 11.91,
          "usd_per_hour": 0.507,
          "usd_per_million": 11.8296,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 2.5199
            },
            {
              "resource": "Memory",
              "quantity": "7.5 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 6.248,
              "usd_per_million": 0.5248
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 8.3998
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.385
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "ZWHEb6YO7EFSsqq3NikNO5847dBPxFOCaQUtvujdik5Tp/hN8XGEoJfU6jKuGdZbHN9f7hyAmMZMwFymBG4qBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "7XR8VyEin+ccoQdtU1Zg2r1GrGL10a6IVI50QOOOHlI="
        },
        "build_page": "/inferences/qwen-3/vl-4b/qwen-f16-ollama",
        "hub_url": "https://huggingface.co/qwen/qwen3-vl-4b-instruct-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": 11.91,
        "context_tokens": 131072,
        "build": "F16 (16-bit, full precision) by Qwen",
        "ref": "hf.co/Qwen/Qwen3-VL-4B-Instruct-GGUF:F16",
        "engine": "ollama-host",
        "cost_usd_per_million": 11.8296,
        "cost_breakdown": {
          "tokens_per_sec": 11.91,
          "usd_per_hour": 0.507,
          "usd_per_million": 11.8296,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 2.5199
            },
            {
              "resource": "Memory",
              "quantity": "7.5 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 6.248,
              "usd_per_million": 0.5248
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 8.3998
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.385
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "ZWHEb6YO7EFSsqq3NikNO5847dBPxFOCaQUtvujdik5Tp/hN8XGEoJfU6jKuGdZbHN9f7hyAmMZMwFymBG4qBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "7XR8VyEin+ccoQdtU1Zg2r1GrGL10a6IVI50QOOOHlI="
        },
        "build_page": "/inferences/qwen-3/vl-4b/qwen-f16-ollama",
        "hub_url": "https://huggingface.co/qwen/qwen3-vl-4b-instruct-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W34",
        "2026-W35",
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 11.91,
          "context_tokens": 131072,
          "build": "F16 (16-bit, full precision) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-VL-4B-Instruct-GGUF:F16",
          "engine": "ollama-host",
          "cost_usd_per_million": 11.8296,
          "cost_breakdown": {
            "tokens_per_sec": 11.91,
            "usd_per_hour": 0.507,
            "usd_per_million": 11.8296,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.5199
              },
              {
                "resource": "Memory",
                "quantity": "7.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 6.248,
                "usd_per_million": 0.5248
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 8.3998
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.385
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "ZWHEb6YO7EFSsqq3NikNO5847dBPxFOCaQUtvujdik5Tp/hN8XGEoJfU6jKuGdZbHN9f7hyAmMZMwFymBG4qBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "7XR8VyEin+ccoQdtU1Zg2r1GrGL10a6IVI50QOOOHlI="
          },
          "build_page": "/inferences/qwen-3/vl-4b/qwen-f16-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-vl-4b-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3-VL-4B-Instruct"
      },
      "builds": 5,
      "variants": [],
      "bottom_line": "12 tokens a second on Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Qwen 3.6 27B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3-6/27b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.6-27B",
      "download_gb": 15,
      "params_b": 27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 5.53,
        "context_tokens": 131072,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3.6-27B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 41.3775,
        "cost_breakdown": {
          "tokens_per_sec": 5.53,
          "usd_per_hour": 0.8235,
          "usd_per_million": 41.3775,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 19.8977
            },
            {
              "resource": "Memory",
              "quantity": "15.7 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 13.052,
              "usd_per_million": 2.361
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 18.0888
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 1.03
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "KsOXom9l/c3ZoV84YzBIIfgMIRYd5X1jhFZRyY3gm7ZA788z5jOt5lOHbSayVZgTrFlHQQ23zApSBMfZPqXdBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "vGPQHnVhlttwDl31XvKYhRNHFw08oGNIOtbhSMsjYr8="
        },
        "build_page": "/inferences/qwen-3-6/27b/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3.6-27b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 5.53,
        "context_tokens": 131072,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3.6-27B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 41.3775,
        "cost_breakdown": {
          "tokens_per_sec": 5.53,
          "usd_per_hour": 0.8235,
          "usd_per_million": 41.3775,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 19.8977
            },
            {
              "resource": "Memory",
              "quantity": "15.7 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 13.052,
              "usd_per_million": 2.361
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 18.0888
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 1.03
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "KsOXom9l/c3ZoV84YzBIIfgMIRYd5X1jhFZRyY3gm7ZA788z5jOt5lOHbSayVZgTrFlHQQ23zApSBMfZPqXdBg==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NS41MjgyNzU4ODMxMDU3NTksInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0MDI4MTkyOTcwOTcyMjA1LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwNTI0MDUwZTkyOTZkMzEyY2QyODI4YzQyNjE2NWJmZjIzYmFlZWY3NjlhZGE5NzA4MzA5ODgzMmYwNmRhZTY3OSIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbnZpZGlhLzY0ZyIsIm1vZGVsIjoiaGYuY28vdW5zbG90aC9Rd2VuMy42LTI3Qi1HR1VGOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo0MCwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjozMDY2LjQxNzE1NCwidG90YWxfdG9rZW5zIjoxNjk1Mn19",
          "payload_sha256": "vGPQHnVhlttwDl31XvKYhRNHFw08oGNIOtbhSMsjYr8="
        },
        "build_page": "/inferences/qwen-3-6/27b/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3.6-27b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 5.53,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3.6-27B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 41.3775,
          "cost_breakdown": {
            "tokens_per_sec": 5.53,
            "usd_per_hour": 0.8235,
            "usd_per_million": 41.3775,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 19.8977
              },
              {
                "resource": "Memory",
                "quantity": "15.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 13.052,
                "usd_per_million": 2.361
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 18.0888
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 1.03
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "KsOXom9l/c3ZoV84YzBIIfgMIRYd5X1jhFZRyY3gm7ZA788z5jOt5lOHbSayVZgTrFlHQQ23zApSBMfZPqXdBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "vGPQHnVhlttwDl31XvKYhRNHFw08oGNIOtbhSMsjYr8="
          },
          "build_page": "/inferences/qwen-3-6/27b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.6-27b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.6-27b",
        "url": "https://openrouter.ai/qwen/qwen3.6-27b",
        "input_usd_per_million": 0.32,
        "output_usd_per_million": 3.2,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.25,
            "usd_per_m_out": 2,
            "context_length": 262144,
            "uptime_1d": 94.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.256,
            "usd_per_m_out": 2.56,
            "context_length": 262144,
            "uptime_1d": null,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 2.7,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 99.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.32,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.325,
            "usd_per_m_out": 3.25,
            "context_length": 256000,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.48,
            "usd_per_m_out": 4.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 6,
            "context_length": 262144,
            "uptime_1d": 98.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 8
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.6-27B"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "5.5 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. About $41 per million tokens here, against OpenRouter's $3.20: the API is 13× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "LFM2.5 24B-A2B",
      "made_by": "LiquidAI",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/61b8e2ba285851687028d395/EsTgVtnM2IqVRKgPdfqcB.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/lfm-2-5/24b-a2b",
      "hub_url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B",
      "download_gb": 12.6,
      "params_b": 24,
      "active_params_b": 2,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 3.64,
        "context_tokens": 4096,
        "build": "IQ3_XXS (3-bit) by unsloth",
        "ref": "hf.co/unsloth/LFM2.5-24B-A2B-GGUF:IQ3_XXS",
        "engine": "ollama-host",
        "cost_usd_per_million": 10.2663,
        "cost_breakdown": {
          "tokens_per_sec": 3.64,
          "usd_per_hour": 0.1345,
          "usd_per_million": 10.2663,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 8.2416
            },
            {
              "resource": "Memory",
              "quantity": "8.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 6.953,
              "usd_per_million": 1.9101
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.1146
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "tYkrlLfxdoF3XBFrQITD54Z8TsIKIZ6w7cOJ/Rf7ozaSpSNvJ9xHImlFCdxXTaK/Vb/rqrcOA2O/uN5RWPlMAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "asYA5ffSMT9zb0bo3p3aQSrFWGP33B0S58oc8IPopfA="
        },
        "build_page": "/inferences/lfm-2-5/24b-a2b/unsloth-iq3_xxs-ollama",
        "hub_url": "https://huggingface.co/unsloth/lfm2.5-24b-a2b-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 3.64,
        "context_tokens": 4096,
        "build": "IQ3_XXS (3-bit) by unsloth",
        "ref": "hf.co/unsloth/LFM2.5-24B-A2B-GGUF:IQ3_XXS",
        "engine": "ollama-host",
        "cost_usd_per_million": 10.2663,
        "cost_breakdown": {
          "tokens_per_sec": 3.64,
          "usd_per_hour": 0.1345,
          "usd_per_million": 10.2663,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 8.2416
            },
            {
              "resource": "Memory",
              "quantity": "8.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 6.953,
              "usd_per_million": 1.9101
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.1146
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "tYkrlLfxdoF3XBFrQITD54Z8TsIKIZ6w7cOJ/Rf7ozaSpSNvJ9xHImlFCdxXTaK/Vb/rqrcOA2O/uN5RWPlMAw==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6My42NDAwODE3MDU5Njc3NjgsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzNDA1OTE3Mjk4NjQ4ODkxLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwZWUzNWRjNDZhMjIxMmNiMzE0NDRjYTRjMTk2NjlhYzhkOTdiMDkyNTIwYjhjOWU0ZDI5YWI0OGQ5ZWU2MWQzYiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbm9uZS8zMmciLCJtb2RlbCI6ImhmLmNvL3Vuc2xvdGgvTEZNMi41LTI0Qi1BMkItR0dVRjpJUTNfWFhTIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjQwLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjMxOTYuOTA2MjYyLCJ0b3RhbF90b2tlbnMiOjExNjM3fX0=",
          "payload_sha256": "asYA5ffSMT9zb0bo3p3aQSrFWGP33B0S58oc8IPopfA="
        },
        "build_page": "/inferences/lfm-2-5/24b-a2b/unsloth-iq3_xxs-ollama",
        "hub_url": "https://huggingface.co/unsloth/lfm2.5-24b-a2b-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 3.64,
          "context_tokens": 4096,
          "build": "IQ3_XXS (3-bit) by unsloth",
          "ref": "hf.co/unsloth/LFM2.5-24B-A2B-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "cost_usd_per_million": 10.2663,
          "cost_breakdown": {
            "tokens_per_sec": 3.64,
            "usd_per_hour": 0.1345,
            "usd_per_million": 10.2663,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 8.2416
              },
              {
                "resource": "Memory",
                "quantity": "8.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 6.953,
                "usd_per_million": 1.9101
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1146
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "tYkrlLfxdoF3XBFrQITD54Z8TsIKIZ6w7cOJ/Rf7ozaSpSNvJ9xHImlFCdxXTaK/Vb/rqrcOA2O/uN5RWPlMAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "asYA5ffSMT9zb0bo3p3aQSrFWGP33B0S58oc8IPopfA="
          },
          "build_page": "/inferences/lfm-2-5/24b-a2b/unsloth-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/lfm2.5-24b-a2b-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "liquid/lfm-2.5-24b-a2b",
        "url": "https://openrouter.ai/liquid/lfm-2.5-24b-a2b",
        "input_usd_per_million": 0.05,
        "output_usd_per_million": 0.7,
        "free_tier": false,
        "context_tokens": 128000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Liquid",
            "tag": "liquid/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.05,
            "usd_per_m_out": 0.7,
            "context_length": 128000,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "LFM1.0 licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B"
      },
      "builds": 4,
      "variants": [
        {
          "model": "LFM2.5-24B-A2B-Uncensored-GGUF",
          "made_by": "AtomicChat",
          "page": "/inferences/lfm-2-5/24b-a2b~atomicchat-uncensored",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "3.6 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. About $10 per million tokens here, against OpenRouter's $0.70: the API is 15× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2",
      "made_by": "yuxinlu1",
      "maker_logo": null,
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Text & chat",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF",
      "download_gb": null,
      "params_b": 12,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "English",
        "Hindi",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu",
        "Amharic",
        "Bengali",
        "Gujarati",
        "Kannada",
        "Malayalam",
        "Dutch",
        "Russian"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": 131072,
        "build": null,
        "ref": "hf.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF:Q3_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-23T17:12:38.055Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/gemma-4/12b~yuxinlu1/yuxinlu1-q3_k_m-ollama",
        "hub_url": "https://huggingface.co/yuxinlu1/gemma-4-12b-agentic-fable5-composer2.5-v2-3.5x-tau2-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 4,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": null,
          "ref": "hf.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF:Q3_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T17:12:38.055Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/12b~yuxinlu1/yuxinlu1-q3_k_m-ollama",
          "hub_url": "https://huggingface.co/yuxinlu1/gemma-4-12b-agentic-fable5-composer2.5-v2-3.5x-tau2-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 4096,
          "build": null,
          "ref": "hf.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-24T17:45:14.364Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/12b~yuxinlu1/yuxinlu1-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/yuxinlu1/gemma-4-12b-agentic-fable5-composer2.5-v2-3.5x-tau2-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": null,
          "ref": "hf.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF:Q6_K",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T14:41:23.604Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/12b~yuxinlu1/yuxinlu1-q6_k-ollama",
          "hub_url": "https://huggingface.co/yuxinlu1/gemma-4-12b-agentic-fable5-composer2.5-v2-3.5x-tau2-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": null,
          "ref": "hf.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T14:55:24.559Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/12b~yuxinlu1/yuxinlu1-q8_0-ollama",
          "hub_url": "https://huggingface.co/yuxinlu1/gemma-4-12b-agentic-fable5-composer2.5-v2-3.5x-tau2-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 4,
      "variants": [],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded."
    },
    {
      "model": "Qwen 3 14B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3/14b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3-14B",
      "download_gb": 8.4,
      "params_b": 14,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 65536,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Malayalam",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": 65536,
        "build": "by qwen",
        "ref": "Qwen/Qwen3-14B-AWQ",
        "engine": "vllm",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-23T18:24:45.561Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/qwen-3/14b/qwen-awq-vllm",
        "hub_url": "https://huggingface.co/qwen/qwen3-14b-awq",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 65536,
          "build": "by qwen",
          "ref": "Qwen/Qwen3-14B-AWQ",
          "engine": "vllm",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T18:24:45.561Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-3/14b/qwen-awq-vllm",
          "hub_url": "https://huggingface.co/qwen/qwen3-14b-awq",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 32768,
          "build": "Q8_0 (8-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-14B-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T21:59:41.480Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-3/14b/qwen-q8_0-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-14b-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3-14b",
        "url": "https://openrouter.ai/qwen/qwen3-14b",
        "input_usd_per_million": 0.12,
        "output_usd_per_million": 0.24,
        "free_tier": false,
        "context_tokens": 40960,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.096,
            "usd_per_m_out": 0.192,
            "context_length": 40960,
            "uptime_1d": null,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "NextBit",
            "tag": "nextbit/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.22,
            "context_length": 40960,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.24,
            "context_length": 40960,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.36,
            "context_length": 40960,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 4
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3-14B"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded. OpenRouter sells it at $0.24 per million output tokens."
    },
    {
      "model": "Qwen 3.6 35B-A3B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3-6/35b-a3b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B",
      "download_gb": 19,
      "params_b": 35,
      "active_params_b": 3,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": 4096,
        "build": "IQ1_M (1-bit) by unsloth",
        "ref": "unsloth/Qwen3.6-35B-A3B-GGUF:IQ1_M",
        "engine": "llamacpp",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "AMD EPYC 7B12",
            "model_raw": "AMD EPYC 7B12",
            "clock_ghz": null,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 512,
            "l3_mb": 16,
            "single_core_hashes_per_sec": 1460.2,
            "multi_core_hashes_per_sec": 3246.1,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T01:38:20.212Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 6.3,
            "latency_ns": 21.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T01:38:21.135Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4982.5,
            "write_mb_s": 179.3,
            "random_read_iops": 476287,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-15T01:38:21.741Z",
            "measured_in": "TW_110",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-08T03:28:28.486Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/qwen-3-6/35b-a3b/unsloth-iq1_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3.6-35b-a3b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "662a4e"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 4096,
          "build": "IQ1_M (1-bit) by unsloth",
          "ref": "unsloth/Qwen3.6-35B-A3B-GGUF:IQ1_M",
          "engine": "llamacpp",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-08T03:28:28.486Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-3-6/35b-a3b/unsloth-iq1_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.6-35b-a3b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.6-35b-a3b",
        "url": "https://openrouter.ai/qwen/qwen3.6-35b-a3b",
        "input_usd_per_million": 0.1,
        "output_usd_per_million": 0.9,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.066,
            "usd_per_m_out": 0.602,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.05,
            "usd_per_m_out": 0.7,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.76,
            "context_length": 262144,
            "uptime_1d": 85.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.89,
            "context_length": 262144,
            "uptime_1d": 91.9,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.9,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.9,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.95,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 1,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 1,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.25,
            "usd_per_m_out": 1.25,
            "context_length": 262144,
            "uptime_1d": 29.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.425,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 1.8,
            "context_length": 262144,
            "uptime_1d": 95.3,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 12
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B"
      },
      "builds": 6,
      "variants": [
        {
          "model": "Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF",
          "made_by": "huihui-ai",
          "page": "/inferences/qwen-3-6/35b-a3b~huihui-ai-abliterated-mtp",
          "status": "measured",
          "tokens_per_sec": 5
        },
        {
          "model": "Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
          "made_by": "llmfan46",
          "page": "/inferences/qwen-3-6/35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved",
          "status": "measured",
          "tokens_per_sec": 4.85
        },
        {
          "model": "Qwen3.6-35B-A3B-BE-GGUF",
          "made_by": "taronaeo",
          "page": "/inferences/qwen-3-6/35b-a3b~taronaeo-be",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded. OpenRouter sells it at $0.90 per million output tokens."
    },
    {
      "model": "MiniCPM5 1B",
      "made_by": "openbmb",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1670387859384-633fe7784b362488336bbfad.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/minicpm-5/1b",
      "hub_url": "https://huggingface.co/openbmb/MiniCPM5-1B-GGUF",
      "download_gb": 0.6,
      "params_b": 1,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 65536,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": "AMD Ryzen 7 5800X 8-Core",
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "GeForce RTX 3080 Ti",
          "gpu_count": 1,
          "vram_gb": 12,
          "host": "own machine",
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
        "machines": 1,
        "tokens_per_sec": 381.36,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM5-1B-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": 0.6733,
        "cost_breakdown": {
          "tokens_per_sec": 381.36,
          "usd_per_hour": 0.9244,
          "usd_per_million": 0.6733,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "15 of 16 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 150,
              "usd_per_million": 0.3933
            },
            {
              "resource": "Memory",
              "quantity": "0.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.534,
              "usd_per_million": 0.0014
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.2622
            },
            {
              "resource": "Energy",
              "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 6.25,
              "usd_per_million": 0.0164
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-10T11:20:52.995Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "ozh3hx0/d7WPcYqg+IkhH50y8z/CTeLYGY4A2j49FKW0nGMHRvC/A+sw3CvrGSTdkiErOw1Lqp9qoKP3Vw6KDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "LnhDHC/Su98gwUes4AKE6RB5M1olge99nDgeKxvyc1o="
        },
        "build_page": "/inferences/minicpm-5/1b/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm5-1b-gguf",
        "weeks": [
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "cheapest": {
        "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": "AMD Ryzen 7 5800X 8-Core",
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "GeForce RTX 3080 Ti",
          "gpu_count": 1,
          "vram_gb": 12,
          "host": "own machine",
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
        "machines": 1,
        "tokens_per_sec": 381.36,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM5-1B-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": 0.6733,
        "cost_breakdown": {
          "tokens_per_sec": 381.36,
          "usd_per_hour": 0.9244,
          "usd_per_million": 0.6733,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "15 of 16 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 150,
              "usd_per_million": 0.3933
            },
            {
              "resource": "Memory",
              "quantity": "0.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.534,
              "usd_per_million": 0.0014
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.2622
            },
            {
              "resource": "Energy",
              "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 6.25,
              "usd_per_million": 0.0164
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-10T11:20:52.995Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "ozh3hx0/d7WPcYqg+IkhH50y8z/CTeLYGY4A2j49FKW0nGMHRvC/A+sw3CvrGSTdkiErOw1Lqp9qoKP3Vw6KDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "LnhDHC/Su98gwUes4AKE6RB5M1olge99nDgeKxvyc1o="
        },
        "build_page": "/inferences/minicpm-5/1b/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm5-1b-gguf",
        "weeks": [
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "machine_kinds": 4,
      "machines_total": 4,
      "weeks_measured": [
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": 381.36,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM5-1B-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 0.6733,
          "cost_breakdown": {
            "tokens_per_sec": 381.36,
            "usd_per_hour": 0.9244,
            "usd_per_million": 0.6733,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.3933
              },
              {
                "resource": "Memory",
                "quantity": "0.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.534,
                "usd_per_million": 0.0014
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.2622
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0164
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "ozh3hx0/d7WPcYqg+IkhH50y8z/CTeLYGY4A2j49FKW0nGMHRvC/A+sw3CvrGSTdkiErOw1Lqp9qoKP3Vw6KDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "LnhDHC/Su98gwUes4AKE6RB5M1olge99nDgeKxvyc1o="
          },
          "build_page": "/inferences/minicpm-5/1b/openbmb-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm5-1b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 208.3,
          "context_tokens": 65536,
          "build": "Q4_K_M (4-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM5-1B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.0381,
          "cost_breakdown": {
            "tokens_per_sec": 208.3,
            "usd_per_hour": 0.7784,
            "usd_per_million": 1.0381,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 0.5281
              },
              {
                "resource": "Memory",
                "quantity": "0.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.534,
                "usd_per_million": 0.0026
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.4801
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0273
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "Oqg8YQNW3f1j701J0bwXWiEJMF3KbfvCKE9GgZ5FtkknN1I52uQUgZA9cfr+3MZS8Je6SEF9vYdbntxOxkJDAQ==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MjA4LjI5Njk0Nzc5MDE1NjU0LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5NDAyODE5Mjk3MDk3MjIwNSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDUyNDA1MGU5Mjk2ZDMxMmNkMjgyOGM0MjYxNjViZmYyM2JhZWVmNzY5YWRhOTcwODMwOTg4MzJmMDZkYWU2NzkiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L252aWRpYS82NGciLCJtb2RlbCI6ImhmLmNvL29wZW5ibWIvTWluaUNQTTUtMUItR0dVRjpRNF9LX00iLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6NDMsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6OTguNjE0MDIyOTk5LCJ0b3RhbF90b2tlbnMiOjIwNTQxfX0=",
            "payload_sha256": "xboKOF4UlUmAdXUmNXxRHX1CuzgXDXoyabPHZo1QVww="
          },
          "build_page": "/inferences/minicpm-5/1b/openbmb-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm5-1b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": 21.73,
          "context_tokens": 8192,
          "build": "Q4_K_M (4-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM5-1B-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 3.2909,
          "cost_breakdown": {
            "tokens_per_sec": 21.73,
            "usd_per_hour": 0.2574,
            "usd_per_million": 3.2909,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 3.2216
              },
              {
                "resource": "Memory",
                "quantity": "0.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.534,
                "usd_per_million": 0.0246
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0447
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "axoj7LjWYgugssNO1lXIf75xcEnquLUZgHdQwZiziQpUeJuDvYzpe3+Op+Uy4fc0YzRKdyxWQ+ICVaD6V22hAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "roAsaoDsgMIV2mA3qhUMZCSnR6/rHTR5v2CdwwEMs5U="
          },
          "build_page": "/inferences/minicpm-5/1b/openbmb-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm5-1b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 20.58,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM5-1B-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 3.475,
          "cost_breakdown": {
            "tokens_per_sec": 20.58,
            "usd_per_hour": 0.2574,
            "usd_per_million": 3.475,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 3.4018
              },
              {
                "resource": "Memory",
                "quantity": "0.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.534,
                "usd_per_million": 0.026
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0472
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "TIaWwHklgMQcq9MaF4I+hUe94cKZonVPoNNbjD4oU32l6Uy8LN9HZ0up9gDuQDnFEi1t3ATcgh1dbvwTJk6cBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "fwi+MDyQRHe+7NL9pYSzGNYC6RpRtz6LaVFxLXmL7nA="
          },
          "build_page": "/inferences/minicpm-5/1b/openbmb-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm5-1b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openbmb/MiniCPM5-1B-GGUF"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "381 tokens a second on Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "MiniCPM-o 4.5 Base",
      "made_by": "openbmb",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1670387859384-633fe7784b362488336bbfad.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/minicpm-o-4-5/base",
      "hub_url": "https://huggingface.co/openbmb/MiniCPM-o-4_5",
      "download_gb": 4.7,
      "params_b": 8.3,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": 8192,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 20.23,
        "context_tokens": 8192,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM-o-4_5-gguf:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 10.8538,
        "cost_breakdown": {
          "tokens_per_sec": 20.23,
          "usd_per_hour": 0.7905,
          "usd_per_million": 10.8538,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 5.4369
            },
            {
              "resource": "Memory",
              "quantity": "4.7 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 3.901,
              "usd_per_million": 0.1928
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 4.9426
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.2814
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "ogwChQwpPuTIlu+sxQDjLCkJYV5a++0D44GKTGOGJSQT0M+Y8l0xDk2TuwyVBA9YSDZZcISLjUjVuC1zWYpzAA==",
          "signing_payload_b64": "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",
          "payload_sha256": "E3L5XMdxEzr/2VHf6aD/uH9RVokArOO2o/yAEAh0vBQ="
        },
        "build_page": "/inferences/minicpm-o-4-5/base/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm-o-4_5-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 20.23,
        "context_tokens": 8192,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM-o-4_5-gguf:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 10.8538,
        "cost_breakdown": {
          "tokens_per_sec": 20.23,
          "usd_per_hour": 0.7905,
          "usd_per_million": 10.8538,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 5.4369
            },
            {
              "resource": "Memory",
              "quantity": "4.7 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 3.901,
              "usd_per_million": 0.1928
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 4.9426
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.2814
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "ogwChQwpPuTIlu+sxQDjLCkJYV5a++0D44GKTGOGJSQT0M+Y8l0xDk2TuwyVBA9YSDZZcISLjUjVuC1zWYpzAA==",
          "signing_payload_b64": "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",
          "payload_sha256": "E3L5XMdxEzr/2VHf6aD/uH9RVokArOO2o/yAEAh0vBQ="
        },
        "build_page": "/inferences/minicpm-o-4-5/base/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm-o-4_5-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 20.23,
          "context_tokens": 8192,
          "build": "Q4_K_M (4-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM-o-4_5-gguf:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 10.8538,
          "cost_breakdown": {
            "tokens_per_sec": 20.23,
            "usd_per_hour": 0.7905,
            "usd_per_million": 10.8538,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 5.4369
              },
              {
                "resource": "Memory",
                "quantity": "4.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.901,
                "usd_per_million": 0.1928
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 4.9426
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.2814
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "ogwChQwpPuTIlu+sxQDjLCkJYV5a++0D44GKTGOGJSQT0M+Y8l0xDk2TuwyVBA9YSDZZcISLjUjVuC1zWYpzAA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MjAuMjMyMDY0MzIxNjkyNDQ1LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5NDAyODE5Mjk3MDk3MjIwNSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDUyNDA1MGU5Mjk2ZDMxMmNkMjgyOGM0MjYxNjViZmYyM2JhZWVmNzY5YWRhOTcwODMwOTg4MzJmMDZkYWU2NzkiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L252aWRpYS82NGciLCJtb2RlbCI6ImhmLmNvL29wZW5ibWIvTWluaUNQTS1vLTRfNS1nZ3VmOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo1Nywic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjo0NDQuMTk1ODk4OTk5OTk5OTQsInRvdGFsX3Rva2VucyI6ODk4N319",
            "payload_sha256": "E3L5XMdxEzr/2VHf6aD/uH9RVokArOO2o/yAEAh0vBQ="
          },
          "build_page": "/inferences/minicpm-o-4-5/base/openbmb-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm-o-4_5-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openbmb/MiniCPM-o-4_5"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "20 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Kimi Linear 48B-A3B",
      "made_by": "moonshotai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/kimi-linear/48b-a3b",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-Linear-48B-A3B-Instruct",
      "download_gb": 28,
      "params_b": 48,
      "active_params_b": 3,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 2,
        "tokens_per_sec": 4.81,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by bartowski",
        "ref": "hf.co/bartowski/moonshotai_Kimi-Linear-48B-A3B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 11.1806,
        "cost_breakdown": {
          "tokens_per_sec": 4.81,
          "usd_per_hour": 0.1935,
          "usd_per_million": 11.1806,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 6.2407
            },
            {
              "resource": "Memory",
              "quantity": "28.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 23.33,
              "usd_per_million": 4.8531
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0867
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "YoELUsrRXI23U/VSXa434Rs3mzV4HdxRxj5kFzfivKCYDZLhtsLqnLZ1a8LlQsgSq1lCJ0CAZd82m9FAFNTfDg==",
          "signing_payload_b64": "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",
          "payload_sha256": "Fus3Neuar8wsWRwwszugDZ3VoBe/dlQ3FA4A63ERgZo="
        },
        "build_page": "/inferences/kimi-linear/48b-a3b/bartowski-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/bartowski/moonshotai_kimi-linear-48b-a3b-instruct-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b",
          "e3654f"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 2,
        "tokens_per_sec": 4.81,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by bartowski",
        "ref": "hf.co/bartowski/moonshotai_Kimi-Linear-48B-A3B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 11.1806,
        "cost_breakdown": {
          "tokens_per_sec": 4.81,
          "usd_per_hour": 0.1935,
          "usd_per_million": 11.1806,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 6.2407
            },
            {
              "resource": "Memory",
              "quantity": "28.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 23.33,
              "usd_per_million": 4.8531
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0867
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "YoELUsrRXI23U/VSXa434Rs3mzV4HdxRxj5kFzfivKCYDZLhtsLqnLZ1a8LlQsgSq1lCJ0CAZd82m9FAFNTfDg==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NC44MDcxODcyNTI4OTA0Mjg1LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5MzQwNTkxNzI5ODY0ODg5MSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMGVlMzVkYzQ2YTIyMTJjYjMxNDQ0Y2E0YzE5NjY5YWM4ZDk3YjA5MjUyMGI4YzllNGQyOWFiNDhkOWVlNjFkM2IiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L25vbmUvMzJnIiwibW9kZWwiOiJoZi5jby9iYXJ0b3dza2kvbW9vbnNob3RhaV9LaW1pLUxpbmVhci00OEItQTNCLUluc3RydWN0LUdHVUY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjQ1LCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjUyOS4yMDc1OTM5OTgwMDAyLCJ0b3RhbF90b2tlbnMiOjI1NDR9fQ==",
          "payload_sha256": "Fus3Neuar8wsWRwwszugDZ3VoBe/dlQ3FA4A63ERgZo="
        },
        "build_page": "/inferences/kimi-linear/48b-a3b/bartowski-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/bartowski/moonshotai_kimi-linear-48b-a3b-instruct-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b",
          "e3654f"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 2,
          "tokens_per_sec": 4.81,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/moonshotai_Kimi-Linear-48B-A3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 11.1806,
          "cost_breakdown": {
            "tokens_per_sec": 4.81,
            "usd_per_hour": 0.1935,
            "usd_per_million": 11.1806,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.2407
              },
              {
                "resource": "Memory",
                "quantity": "28.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 23.33,
                "usd_per_million": 4.8531
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0867
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "YoELUsrRXI23U/VSXa434Rs3mzV4HdxRxj5kFzfivKCYDZLhtsLqnLZ1a8LlQsgSq1lCJ0CAZd82m9FAFNTfDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "Fus3Neuar8wsWRwwszugDZ3VoBe/dlQ3FA4A63ERgZo="
          },
          "build_page": "/inferences/kimi-linear/48b-a3b/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/moonshotai_kimi-linear-48b-a3b-instruct-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b",
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/moonshotai/Kimi-Linear-48B-A3B-Instruct"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "4.8 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "North 1.0 Mini-Code-1-0",
      "made_by": "CohereLabs",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1678549441248-5e70f6048ce3c604d78fe133.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/north-1/mini-code-1-0",
      "hub_url": "https://huggingface.co/CohereLabs/North-Mini-Code-1.0",
      "download_gb": 17.2,
      "params_b": 31,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 2,
        "tokens_per_sec": 3.2,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/North-Mini-Code-1.0-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 14.1577,
        "cost_breakdown": {
          "tokens_per_sec": 3.2,
          "usd_per_hour": 0.1632,
          "usd_per_million": 14.1577,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 9.3719
            },
            {
              "resource": "Memory",
              "quantity": "17.9 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 14.903,
              "usd_per_million": 4.6556
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.1303
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "4NB1Ukny6rhc1gY5u7446P3vFgG6oB9moRtcbcbWfww4UJOlNqUWfOT/66qELlEB+cwYmROmGNEnDD8wVrq1Cw==",
          "signing_payload_b64": "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",
          "payload_sha256": "RUJWzDwzT1IdZg+m2HmpV1AKoMLpujT0k3JPev+qVZc="
        },
        "build_page": "/inferences/north-1/mini-code-1-0/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/north-mini-code-1.0-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b",
          "e3654f"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 2,
        "tokens_per_sec": 3.2,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/North-Mini-Code-1.0-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 14.1577,
        "cost_breakdown": {
          "tokens_per_sec": 3.2,
          "usd_per_hour": 0.1632,
          "usd_per_million": 14.1577,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 9.3719
            },
            {
              "resource": "Memory",
              "quantity": "17.9 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 14.903,
              "usd_per_million": 4.6556
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.1303
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "4NB1Ukny6rhc1gY5u7446P3vFgG6oB9moRtcbcbWfww4UJOlNqUWfOT/66qELlEB+cwYmROmGNEnDD8wVrq1Cw==",
          "signing_payload_b64": "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",
          "payload_sha256": "RUJWzDwzT1IdZg+m2HmpV1AKoMLpujT0k3JPev+qVZc="
        },
        "build_page": "/inferences/north-1/mini-code-1-0/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/north-mini-code-1.0-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b",
          "e3654f"
        ]
      },
      "machine_kinds": 2,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W44",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 2,
          "tokens_per_sec": 3.2,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/North-Mini-Code-1.0-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 14.1577,
          "cost_breakdown": {
            "tokens_per_sec": 3.2,
            "usd_per_hour": 0.1632,
            "usd_per_million": 14.1577,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 9.3719
              },
              {
                "resource": "Memory",
                "quantity": "17.9 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 14.903,
                "usd_per_million": 4.6556
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1303
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "4NB1Ukny6rhc1gY5u7446P3vFgG6oB9moRtcbcbWfww4UJOlNqUWfOT/66qELlEB+cwYmROmGNEnDD8wVrq1Cw==",
            "signing_payload_b64": "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",
            "payload_sha256": "RUJWzDwzT1IdZg+m2HmpV1AKoMLpujT0k3JPev+qVZc="
          },
          "build_page": "/inferences/north-1/mini-code-1-0/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/north-mini-code-1.0-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b",
            "e3654f"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 2.15,
          "context_tokens": 4096,
          "build": "IQ3_XXS (3-bit) by mradermacher",
          "ref": "hf.co/mradermacher/North-Mini-Code-1.0-i1-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "cost_usd_per_million": 18.4585,
          "cost_breakdown": {
            "tokens_per_sec": 2.15,
            "usd_per_hour": 0.1427,
            "usd_per_million": 18.4585,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 13.9682
              },
              {
                "resource": "Memory",
                "quantity": "11.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 9.227,
                "usd_per_million": 4.2962
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1942
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:44:44.996Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "mvXYbGmkkpE3fb3bJmFTimbVH7pWkgYZ4KmlSNmkPQ1LvbtYcT8uh/KmCpsJxYKPcVHJRGX7wNKZ0gT4j+ppDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "iqJSHOJAuxFLESHFD5j/W7RIfZUN+hkWG4NRGrmbow4="
          },
          "build_page": "/inferences/north-1/mini-code-1-0~mradermacher-i1/mradermacher-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/mradermacher/north-mini-code-1.0-i1-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/CohereLabs/North-Mini-Code-1.0"
      },
      "builds": 5,
      "variants": [],
      "bottom_line": "3.2 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP-GGUF",
      "made_by": "HauhauCS",
      "maker_logo": "https://huggingface.co/avatars/1a732618ac3a62f81515f76145c902a6.svg",
      "based_on": "Qwen 3.8 27B",
      "variant_of": "/inferences/qwen-3-8/27b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/qwen-3-8/27b~hauhaucs-uncensored-aggressive-mtp",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "download_gb": 24.1,
      "params_b": 27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 65536,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "machine not reported",
        "machine_detail": null,
        "machine_class": "unknown",
        "machines": 1,
        "tokens_per_sec": 1.75,
        "context_tokens": 65536,
        "build": "Q6_K_P (6-bit) by HauhauCS",
        "ref": "hf.co/HauhauCS/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP-GGUF:Q6_K_P",
        "engine": "ollama",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "4N4JLpP8Ih/qE3Dp3fyNdxdDgN/1G5vp/AL8Is22PHkX842BuutWc77Hx2iNZNsLLAHS11kYTW49kinSNVUzCA==",
          "signing_payload_b64": "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",
          "payload_sha256": "XgyhbxXqAxhc034fvYmVMaFK6+b7s2ThQeN4nlWPMg0="
        },
        "build_page": "/inferences/qwen-3-8/27b~hauhaucs-uncensored-aggressive-mtp/hauhaucs-q6_k_p-ollama",
        "hub_url": "https://huggingface.co/hauhaucs/qwen3.8-27b-uncensored-hauhaucs-aggressive-mtp-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "machine not reported",
          "machine_detail": null,
          "machine_class": "unknown",
          "machines": 1,
          "tokens_per_sec": 1.75,
          "context_tokens": 65536,
          "build": "Q6_K_P (6-bit) by HauhauCS",
          "ref": "hf.co/HauhauCS/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP-GGUF:Q6_K_P",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "4N4JLpP8Ih/qE3Dp3fyNdxdDgN/1G5vp/AL8Is22PHkX842BuutWc77Hx2iNZNsLLAHS11kYTW49kinSNVUzCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "XgyhbxXqAxhc034fvYmVMaFK6+b7s2ThQeN4nlWPMg0="
          },
          "build_page": "/inferences/qwen-3-8/27b~hauhaucs-uncensored-aggressive-mtp/hauhaucs-q6_k_p-ollama",
          "hub_url": "https://huggingface.co/hauhaucs/qwen3.8-27b-uncensored-hauhaucs-aggressive-mtp-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.8-27b",
        "url": "https://openrouter.ai/qwen/qwen3.8-27b",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 1.875,
        "free_tier": false,
        "context_tokens": 1024000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.11,
            "usd_per_m_out": 0.781,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake",
            "quantization": null,
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 1.05,
            "context_length": 1024000,
            "uptime_1d": 97.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.039,
            "usd_per_m_out": 1.25,
            "context_length": 262144,
            "uptime_1d": 91.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 1.35,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cerebras",
            "tag": "cerebras/fp16",
            "quantization": "fp16",
            "usd_per_m_in": 0.99,
            "usd_per_m_out": 1.49,
            "context_length": 65536,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.035,
            "usd_per_m_out": 1.5,
            "context_length": 262144,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Tenstorrent",
            "tag": "tenstorrent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.6,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.875,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 1.95,
            "context_length": 262144,
            "uptime_1d": 91.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 1.98,
            "context_length": 262144,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.125,
            "usd_per_m_out": 2,
            "context_length": 262144,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 2.2,
            "context_length": 262144,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 2.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 2.3,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.425,
            "usd_per_m_out": 2.55,
            "context_length": 1000000,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.4,
            "usd_per_m_out": 3,
            "context_length": 262144,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.42,
            "usd_per_m_out": 3,
            "context_length": 1000000,
            "uptime_1d": 79.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 95.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 23
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.8-27B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "1.8 tokens a second on a machine not reported. OpenRouter sells it at $1.88 per million output tokens."
    },
    {
      "model": "Ministral 3 14B-Instruct-2512",
      "made_by": "mistralai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/634c17653d11eaedd88b314d/9OgyfKstSZtbmsmuG8MbU.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/ministral-3/14b-instruct-2512",
      "hub_url": "https://huggingface.co/mistralai/Ministral-3-14B-Instruct-2512",
      "download_gb": 6.7,
      "params_b": 14,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by mistralai",
        "ref": "hf.co/mistralai/Ministral-3-14B-Instruct-2512-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T01:50:29.183Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/ministral-3/14b-instruct-2512/mistralai-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/mistralai/ministral-3-14b-instruct-2512-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by mistralai",
          "ref": "hf.co/mistralai/Ministral-3-14B-Instruct-2512-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T01:50:29.183Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/ministral-3/14b-instruct-2512/mistralai-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/mistralai/ministral-3-14b-instruct-2512-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "mistralai/ministral-14b-2512",
        "url": "https://openrouter.ai/mistralai/ministral-14b-2512",
        "input_usd_per_million": 0.2,
        "output_usd_per_million": 0.2,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": null,
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 0.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 0.22,
            "usd_per_m_out": 0.22,
            "context_length": 262144,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/mistralai/Ministral-3-14B-Instruct-2512"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded. OpenRouter sells it at $0.20 per million output tokens."
    },
    {
      "model": "Qwen 3 1.7B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3/1.7b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3-1.7B",
      "download_gb": null,
      "params_b": 1.7,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 65536,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": 65536,
        "build": "by qwen",
        "ref": "Qwen/Qwen3-1.7B",
        "engine": "vllm",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-23T19:52:27.499Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/qwen-3/1.7b/qwen-bf16-vllm",
        "hub_url": "https://huggingface.co/qwen/qwen3-1.7b",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 65536,
          "build": "by qwen",
          "ref": "Qwen/Qwen3-1.7B",
          "engine": "vllm",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T19:52:27.499Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-3/1.7b/qwen-bf16-vllm",
          "hub_url": "https://huggingface.co/qwen/qwen3-1.7b",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3-1.7B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded."
    },
    {
      "model": "Qwen 3 8B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3/8b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3-8B",
      "download_gb": null,
      "params_b": 8,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 65536,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": 65536,
        "build": "by qwen",
        "ref": "Qwen/Qwen3-8B",
        "engine": "vllm",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-23T20:45:46.870Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/qwen-3/8b/qwen-bf16-vllm",
        "hub_url": "https://huggingface.co/qwen/qwen3-8b",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 65536,
          "build": "by qwen",
          "ref": "Qwen/Qwen3-8B",
          "engine": "vllm",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T20:45:46.870Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-3/8b/qwen-bf16-vllm",
          "hub_url": "https://huggingface.co/qwen/qwen3-8b",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3-8B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded."
    },
    {
      "model": "Llama 3.2 3B",
      "made_by": "meta-llama",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/646cf8084eefb026fb8fd8bc/oCTqufkdTkjyGodsx1vo1.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/llama-3-2/3b",
      "hub_url": "https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct",
      "download_gb": 1.9,
      "params_b": 3.2,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "English",
        "Malayalam",
        "Telugu",
        "Hindi",
        "Kannada",
        "Bengali",
        "Marathi",
        "Punjabi",
        "Tamil"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 12,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|||"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|||",
        "machines": 1,
        "tokens_per_sec": 225.19,
        "context_tokens": null,
        "build": "IQ1_S (1-bit) by unsloth",
        "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ1_S",
        "engine": "ollama",
        "cost_usd_per_million": 1.1411,
        "cost_breakdown": {
          "tokens_per_sec": 225.19,
          "usd_per_hour": 0.925,
          "usd_per_million": 1.1411,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "15 of 16 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 150,
              "usd_per_million": 0.6661
            },
            {
              "resource": "Memory",
              "quantity": "0.8 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.708,
              "usd_per_million": 0.0031
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.4441
            },
            {
              "resource": "Energy",
              "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 6.25,
              "usd_per_million": 0.0278
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-10T11:20:52.995Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "PveBSNBhE0lObW4aP806WVjqea6j8d4LxYTNYJmQWaKaIPSAns5BY2lQrmwh0hVRGSUW1mbi9QGmp9L1kdYoDQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "79WvM1JAceTQ8k+h2yLBG41bLAwQYkVveC5r27WpGzM="
        },
        "build_page": "/inferences/llama-3-2/3b/unsloth-iq1_s-ollama",
        "hub_url": "https://huggingface.co/unsloth/llama-3.2-3b-instruct-gguf",
        "weeks": [
          "2026-W38",
          "2026-W43"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "cheapest": {
        "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 12,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|||"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|||",
        "machines": 1,
        "tokens_per_sec": 225.19,
        "context_tokens": null,
        "build": "IQ1_S (1-bit) by unsloth",
        "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ1_S",
        "engine": "ollama",
        "cost_usd_per_million": 1.1411,
        "cost_breakdown": {
          "tokens_per_sec": 225.19,
          "usd_per_hour": 0.925,
          "usd_per_million": 1.1411,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "15 of 16 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 150,
              "usd_per_million": 0.6661
            },
            {
              "resource": "Memory",
              "quantity": "0.8 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.708,
              "usd_per_million": 0.0031
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.4441
            },
            {
              "resource": "Energy",
              "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 6.25,
              "usd_per_million": 0.0278
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-10T11:20:52.995Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "PveBSNBhE0lObW4aP806WVjqea6j8d4LxYTNYJmQWaKaIPSAns5BY2lQrmwh0hVRGSUW1mbi9QGmp9L1kdYoDQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "79WvM1JAceTQ8k+h2yLBG41bLAwQYkVveC5r27WpGzM="
        },
        "build_page": "/inferences/llama-3-2/3b/unsloth-iq1_s-ollama",
        "hub_url": "https://huggingface.co/unsloth/llama-3.2-3b-instruct-gguf",
        "weeks": [
          "2026-W38",
          "2026-W43"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "machine_kinds": 5,
      "machines_total": 17,
      "weeks_measured": [
        "2026-W35",
        "2026-W38",
        "2026-W43",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 225.19,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by unsloth",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ1_S",
          "engine": "ollama",
          "cost_usd_per_million": 1.1411,
          "cost_breakdown": {
            "tokens_per_sec": 225.19,
            "usd_per_hour": 0.925,
            "usd_per_million": 1.1411,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.6661
              },
              {
                "resource": "Memory",
                "quantity": "0.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.708,
                "usd_per_million": 0.0031
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.4441
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0278
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "PveBSNBhE0lObW4aP806WVjqea6j8d4LxYTNYJmQWaKaIPSAns5BY2lQrmwh0hVRGSUW1mbi9QGmp9L1kdYoDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "79WvM1JAceTQ8k+h2yLBG41bLAwQYkVveC5r27WpGzM="
          },
          "build_page": "/inferences/llama-3-2/3b/unsloth-iq1_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 205.06,
          "context_tokens": null,
          "build": "Q4_0 (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_0",
          "engine": "ollama",
          "cost_usd_per_million": 1.2569,
          "cost_breakdown": {
            "tokens_per_sec": 205.06,
            "usd_per_hour": 0.9279,
            "usd_per_million": 1.2569,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.7315
              },
              {
                "resource": "Memory",
                "quantity": "1.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.491,
                "usd_per_million": 0.0073
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.4877
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0305
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "N6VJZ0GRJGfittPj1S+QlKfDca2DsANkEqbsFu5oiR1QDyUs6azVaTBr90Kku1XEjCwY4CnkIK5quG1IEpLoBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "2vdc8pjZzyGvjZS3dZwlxnchlSNWNEIXaOhBMHlLIGM="
          },
          "build_page": "/inferences/llama-3-2/3b/unsloth-q4_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W38"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 203.94,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 1.2642,
          "cost_breakdown": {
            "tokens_per_sec": 203.94,
            "usd_per_hour": 0.9281,
            "usd_per_million": 1.2642,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.7355
              },
              {
                "resource": "Memory",
                "quantity": "1.9 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.567,
                "usd_per_million": 0.0077
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.4903
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0306
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "t8ixuFktyclVNDUZo4dKyZBjnZGFFh+UoLFgCcDoy3OQXL6wfZ/gCPvIDFFKKaR1d4irKOdUHYTcRBHtTBNTBQ==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MjAzLjk0MjQ1NTg1NTYzMDMzLCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5NDMwOTY1Mjk5NTAzMjEwMCwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMGJjNjUwYjU5YWY4Nzk0ODk2MDMxZDJlODJiY2EwZTExOWYzOTBlMjFkNjkwOTYzYWU0NzU3ODIzODA4NGJmNTQiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6IndpbmRvd3MvYW1kNjQvbnZpZGlhLzE2ZyIsIm1vZGVsIjoiaGYuY28vdW5zbG90aC9MbGFtYS0zLjItM0ItSW5zdHJ1Y3QtR0dVRjpRNF9LX00iLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEiLCJzYW1wbGVfY291bnQiOjEwLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjAuOTI2NzMyLCJ0b3RhbF90b2tlbnMiOjE4OX19",
            "payload_sha256": "q31wWJLyb4Pqc9fAEUzKmkn0Kro2XXMp8qspXtnZL4I="
          },
          "build_page": "/inferences/llama-3-2/3b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W38",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 199.51,
          "context_tokens": null,
          "build": "IQ3_XXS (3-bit) by unsloth",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ3_XXS",
          "engine": "ollama",
          "cost_usd_per_million": 1.2897,
          "cost_breakdown": {
            "tokens_per_sec": 199.51,
            "usd_per_hour": 0.9263,
            "usd_per_million": 1.2897,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.7518
              },
              {
                "resource": "Memory",
                "quantity": "1.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.063,
                "usd_per_million": 0.0053
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.5012
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0313
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "trRpa25ZFobEaCmnWiOPXmY4losN8RnC5CYtyRivoa22T+qICZHrHATti3MLoR+Ii1gTliDDINjwsSjgWguqAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "ueKoMj7B5IY4inlr3nWTBw0zc3UePxtpUHareQKKDvQ="
          },
          "build_page": "/inferences/llama-3-2/3b/unsloth-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W38"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 124.17,
          "context_tokens": null,
          "build": "IQ2_XXS (2-bit) by unsloth",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ2_XXS",
          "engine": "ollama",
          "cost_usd_per_million": 2.0703,
          "cost_breakdown": {
            "tokens_per_sec": 124.17,
            "usd_per_hour": 0.9254,
            "usd_per_million": 2.0703,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 1.208
              },
              {
                "resource": "Memory",
                "quantity": "1.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.812,
                "usd_per_million": 0.0065
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.8054
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0503
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "D1ig34cA+0j2p63F5Jbd5y39GCd2BIK4p4MB4BmuSPcdbCUzQYPToHwaQiNF9NOP0ogzadbVJlkuH7p6CWxpCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "bX24ghwlBmpnfjqNC+tb2aPXZYqPWxMItJ2K7jkRRiE="
          },
          "build_page": "/inferences/llama-3-2/3b/unsloth-iq2_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W38"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": 99.3,
          "context_tokens": 65536,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 2.5805,
          "cost_breakdown": {
            "tokens_per_sec": 99.3,
            "usd_per_hour": 0.9225,
            "usd_per_million": 2.5805,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 1.5105
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.007
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0629
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "EGkn4eE+egBD8Zqg5buqWFG8Tq93+IMRRvCyo9MiBbVqt/RW22TI3Do7s93q/2TeilF35Mf0fW8qlHg/K4CQDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "ZJ1pHvVEwj/sL1YYDZXINiwf+RBVhKfITOZC7/TBTlw="
          },
          "build_page": "/inferences/llama-3-2/3b/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 97.41,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.2303,
          "cost_breakdown": {
            "tokens_per_sec": 97.41,
            "usd_per_hour": 0.7821,
            "usd_per_million": 2.2303,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 1.1292
              },
              {
                "resource": "Memory",
                "quantity": "1.9 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.567,
                "usd_per_million": 0.0161
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.0266
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0585
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "UUp1dqpyvr1/PNTEE47OUMP7x4DpkyQZwkT0rpFYMZLlDpxlJ97SNXGq37Lz8dUn0a/r8YgZZ3BB3Q6HKA/jBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "e+QldB5nxZZXzSNtgbc2pv7T4q7lF1tzpI4QKaybOjA="
          },
          "build_page": "/inferences/llama-3-2/3b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W38",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 94.89,
          "context_tokens": 16384,
          "build": "F16 (16-bit, full precision) by unsloth",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:F16",
          "engine": "ollama",
          "cost_usd_per_million": 2.7532,
          "cost_breakdown": {
            "tokens_per_sec": 94.89,
            "usd_per_hour": 0.9405,
            "usd_per_million": 2.7532,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 1.5808
              },
              {
                "resource": "Memory",
                "quantity": "6.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 4.993,
                "usd_per_million": 0.0526
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.0539
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0659
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "q+/oK2hh3prtgVAG8mxMI7wphBElRTpMWoKKyd+XlDDnRPW7wh//67maA9Ym/1bB9bnZyrNh+eFqhXp0zzA8AQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "i+YMgejgL3yOJlpD1XIk2TnJkGAF+WrUX9RwaU43YSQ="
          },
          "build_page": "/inferences/llama-3-2/3b/unsloth-f16-ollama",
          "hub_url": "https://huggingface.co/unsloth/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W38"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 94.46,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by unsloth",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ1_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.4323,
          "cost_breakdown": {
            "tokens_per_sec": 94.46,
            "usd_per_hour": 0.487,
            "usd_per_million": 1.4323,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.3176
              },
              {
                "resource": "Memory",
                "quantity": "0.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.708,
                "usd_per_million": 0.0075
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.0587
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0485
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "1r8Qo78XIyntSJ4wux1aO10CbPqXylxbsVuuHVOb0MFYTyWCIkHpklSihye8QwdRgS/m8462q/NOHVPTZgZUCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "j9OBNkhFbWUVRyVCyHbsiHTg5IskTEA5krZMxk68qv4="
          },
          "build_page": "/inferences/llama-3-2/3b/unsloth-iq1_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 86.87,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.5492,
          "cost_breakdown": {
            "tokens_per_sec": 86.87,
            "usd_per_hour": 0.4845,
            "usd_per_million": 1.5492,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.3453
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.1511
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0528
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "2deLKLJAE4JjduhZthR6zTfgy15tcK+Gt7LAYCIIUJ7Lzet3JHLTG4sdzOu0sSMwDC2ZWRrUzxcEOdSqKrObCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "DTtLVgvXJp5zJ9a/UxKKA3y1le+derUkKmQ0zq8nicY="
          },
          "build_page": "/inferences/llama-3-2/3b/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 78.82,
          "context_tokens": 131072,
          "build": "Q3_K_L (3-bit) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.7074,
          "cost_breakdown": {
            "tokens_per_sec": 78.82,
            "usd_per_hour": 0.4845,
            "usd_per_million": 1.7074,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.3806
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.2686
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0581
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "Wt6aa3sVC0haNDXyaZhxJWhYywD+k2aUhljsN52ngYwj27k5jMUzGDhizfPwzzBtCQw7cNIGGi1ZTxolT5lcCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "6P8cmNtxgglLJ0+2GlldMVkOIVOAm3S/r676PbCx2fM="
          },
          "build_page": "/inferences/llama-3-2/3b/bartowski-q3_k_l-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": 78.03,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 3.284,
          "cost_breakdown": {
            "tokens_per_sec": 78.03,
            "usd_per_hour": 0.9225,
            "usd_per_million": 3.284,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 1.9223
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.2815
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0801
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "u2VgH5kp3E/E7Yl5Y3fWzcYTUrWFr+lqo5EVgo1z9aqtg1niQIlOT5SHK21M0X44te90Pq0D0BQpefWNIYN0CQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "vKBZGRzGOo1PUiaS5jFBPhzIW2tOen8wY4lEWWyKlpc="
          },
          "build_page": null,
          "hub_url": "https://huggingface.co/maziyarpanahi/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 72.81,
          "context_tokens": null,
          "build": "Q8_0 (8-bit) by unsloth",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q8_0",
          "engine": "ollama",
          "cost_usd_per_million": 3.5559,
          "cost_breakdown": {
            "tokens_per_sec": 72.81,
            "usd_per_hour": 0.9321,
            "usd_per_million": 3.5559,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 2.0602
              },
              {
                "resource": "Memory",
                "quantity": "3.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 2.655,
                "usd_per_million": 0.0365
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.3734
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0858
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "go91DeLQ6idWbHdt213XRm0c8EDfjU/0nfTfS9aSqeP0kF58LdDFSzbMS64CeU/SA3K+ElxpdTIrtLJub/GmDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "+YAK7oWyWELAwxYOFPwnqJCzagwu2SjDvJz2qz7IN/E="
          },
          "build_page": "/inferences/llama-3-2/3b/unsloth-q8_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W38"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 71.97,
          "context_tokens": 131072,
          "build": "IQ3_M (3-bit) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:IQ3_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.87,
          "cost_breakdown": {
            "tokens_per_sec": 71.97,
            "usd_per_hour": 0.4845,
            "usd_per_million": 1.87,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.4168
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.3895
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0637
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "y0C6GGftevjyABlHXofX+avXNsuSI7vPoLJHdsLPYF9sPXjQop28u5A4BIDCHNey6ljOZ0IroRIP54bAgZwDAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "89nLsZVL7Wh9Nppa07uqoVObsKhGvYx/hA/YnmkMiT4="
          },
          "build_page": "/inferences/llama-3-2/3b/bartowski-iq3_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": 5.03,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 14.4133,
          "cost_breakdown": {
            "tokens_per_sec": 5.03,
            "usd_per_hour": 0.2611,
            "usd_per_million": 14.4133,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 13.9088
              },
              {
                "resource": "Memory",
                "quantity": "1.9 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.567,
                "usd_per_million": 0.3114
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.1931
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "GL9Os9TZx4O7LCb/cOIjuGxtDw4mfdMIdVpPNwwMJ0/VDkm4WdOC9wesPm8EbIVC3y648rsZAthcVUEzUJJRBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "si2oJrhjr04zYCCKf9m16p0R77k8ZYclC4KtWepKxqg="
          },
          "build_page": "/inferences/llama-3-2/3b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W38",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "F16 (16-bit, full precision) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:F16",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-21T22:57:27.099Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/llama-3-2/3b/bartowski-f16-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q8_0 (8-bit) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-21T22:29:12.561Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/llama-3-2/3b/bartowski-q8_0-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-3b-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "meta-llama/llama-3.2-3b-instruct",
        "url": "https://openrouter.ai/meta-llama/llama-3.2-3b-instruct",
        "input_usd_per_million": 0.0509,
        "output_usd_per_million": 0.335,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Parasail",
            "tag": "parasail/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.05,
            "usd_per_m_out": 0.33,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.0509,
            "usd_per_m_out": 0.335,
            "context_length": 80000,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Llama 3.2 community licence",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct"
      },
      "builds": 17,
      "variants": [
        {
          "model": "Llama-3.2-3B-Instruct-abliterated-GGUF",
          "made_by": "QuantFactory",
          "page": "/inferences/llama-3-2/3b~quantfactory-llama-3-2-abliterated",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "Llama-3.2-3B-Instruct-uncensored",
          "made_by": "chuanli11",
          "page": "/inferences/llama-3-2/3b~chuanli11-llama-3-2-uncensored",
          "status": "measured",
          "tokens_per_sec": 46.8
        }
      ],
      "bottom_line": "225 tokens a second on Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM. About $1.14 per million tokens here, against OpenRouter's $0.34: the API is 3.4× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "MiniCPM-V 4.7 1B",
      "made_by": "openbmb",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1670387859384-633fe7784b362488336bbfad.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/minicpm-v-4-7/1b",
      "hub_url": "https://huggingface.co/openbmb/MiniCPM-V-4.7-1B",
      "download_gb": 0.6,
      "params_b": 1,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 262144,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "returns valid JSON",
        "long answers",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Tamil"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 221.15,
        "context_tokens": 262144,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.9777,
        "cost_breakdown": {
          "tokens_per_sec": 221.15,
          "usd_per_hour": 0.7784,
          "usd_per_million": 0.9777,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 0.4974
            },
            {
              "resource": "Memory",
              "quantity": "0.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.521,
              "usd_per_million": 0.0024
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.4522
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0257
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "nt3XDnfvCaYE1DSM78DqccE1gTM/Inf1U1d3NVrcFbMdzA5EFUPa1yxfBPNoeYQbsGEwEDxA3hcbDM2Qw54BAg==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MjIxLjE0NzMxOTk4MDMzMDI4LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5NDAyODE5Mjk3MDk3MjIwNSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDUyNDA1MGU5Mjk2ZDMxMmNkMjgyOGM0MjYxNjViZmYyM2JhZWVmNzY5YWRhOTcwODMwOTg4MzJmMDZkYWU2NzkiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L252aWRpYS82NGciLCJtb2RlbCI6ImhmLmNvL29wZW5ibWIvTWluaUNQTS1WLTQuNy0xQi1nZ3VmOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo2MSwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjoxMC40NjgxMzQ5OTgsInRvdGFsX3Rva2VucyI6MjMxNX19",
          "payload_sha256": "OVOtN858raplNZexGPF/dG6nWWbKvgPPlaz5RItKc54="
        },
        "build_page": "/inferences/minicpm-v-4-7/1b/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm-v-4.7-1b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 221.15,
        "context_tokens": 262144,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.9777,
        "cost_breakdown": {
          "tokens_per_sec": 221.15,
          "usd_per_hour": 0.7784,
          "usd_per_million": 0.9777,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 0.4974
            },
            {
              "resource": "Memory",
              "quantity": "0.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.521,
              "usd_per_million": 0.0024
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.4522
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0257
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "nt3XDnfvCaYE1DSM78DqccE1gTM/Inf1U1d3NVrcFbMdzA5EFUPa1yxfBPNoeYQbsGEwEDxA3hcbDM2Qw54BAg==",
          "signing_payload_b64": "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",
          "payload_sha256": "OVOtN858raplNZexGPF/dG6nWWbKvgPPlaz5RItKc54="
        },
        "build_page": "/inferences/minicpm-v-4-7/1b/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm-v-4.7-1b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 2,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 221.15,
          "context_tokens": 262144,
          "build": "Q4_K_M (4-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9777,
          "cost_breakdown": {
            "tokens_per_sec": 221.15,
            "usd_per_hour": 0.7784,
            "usd_per_million": 0.9777,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 0.4974
              },
              {
                "resource": "Memory",
                "quantity": "0.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.521,
                "usd_per_million": 0.0024
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.4522
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0257
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "nt3XDnfvCaYE1DSM78DqccE1gTM/Inf1U1d3NVrcFbMdzA5EFUPa1yxfBPNoeYQbsGEwEDxA3hcbDM2Qw54BAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "OVOtN858raplNZexGPF/dG6nWWbKvgPPlaz5RItKc54="
          },
          "build_page": "/inferences/minicpm-v-4-7/1b/openbmb-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm-v-4.7-1b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 185.86,
          "context_tokens": 262144,
          "build": "Q8_0 (8-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.1651,
          "cost_breakdown": {
            "tokens_per_sec": 185.86,
            "usd_per_hour": 0.7795,
            "usd_per_million": 1.1651,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 0.5919
              },
              {
                "resource": "Memory",
                "quantity": "1.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.839,
                "usd_per_million": 0.0045
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.5381
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0306
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "tGG/WQL7bWmpumRYSXrjkTCZH6N+pMvcOaoM0ZngBAUDQBeEZf93nKsEL/ELJOAME+TVJqCjhR7wu3xyNuBPAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "XfQVj5GMlUEiWg+ouAjWE29Nu8Nb2J8UIPiY3+hZ7UE="
          },
          "build_page": "/inferences/minicpm-v-4-7/1b/openbmb-q8_0-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm-v-4.7-1b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 14.97,
          "context_tokens": 8192,
          "build": "Q8_0 (8-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.0879,
          "cost_breakdown": {
            "tokens_per_sec": 14.97,
            "usd_per_hour": 0.1125,
            "usd_per_million": 2.0879,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.004
              },
              {
                "resource": "Memory",
                "quantity": "1.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.839,
                "usd_per_million": 0.056
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0279
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 185.9,
              "multi_core_hashes_per_sec": 588,
              "sustained_ratio": 1.4,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-14T19:37:12.663Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 1.7,
              "latency_ns": 71.3,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-14T19:37:15.832Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2202.4,
              "write_mb_s": 15.5,
              "random_read_iops": 196685,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-14T19:37:24.078Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T20:22:46.948Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "DDoNBNp5bGsCF3kCmz17IPxxsHpmPfkPrgAtP/ADhLj4mCJ3PexOIhPVGr4HKAl6UzfYDivFdH3eO/DCZBeSCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "3JxU2y7bmdoj2i4Y1ZQOpAqkqY0GqMgjh+k8ZVXqeu8="
          },
          "build_page": "/inferences/minicpm-v-4-7/1b/openbmb-q8_0-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm-v-4.7-1b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "4e94b2"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openbmb/MiniCPM-V-4.7-1B"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "221 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "GLM-OCR Base",
      "made_by": "zai-org",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/glm-ocr/base",
      "hub_url": "https://huggingface.co/zai-org/GLM-OCR",
      "download_gb": 0.5,
      "params_b": 0.9,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "reasons about code",
        "returns valid JSON",
        "long answers",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 106.1,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by mradermacher",
        "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.037,
        "cost_breakdown": {
          "tokens_per_sec": 106.1,
          "usd_per_hour": 0.778,
          "usd_per_million": 2.037,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 1.0368
            },
            {
              "resource": "Memory",
              "quantity": "0.5 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.426,
              "usd_per_million": 0.004
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.9425
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0537
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "V/tN/jFjOUBl+pYQsVxuIr+guLrtf7ineA8t0YZRhoiqFy2pfa+VQvCs/oEFBkBygoaEb53lY4MZGjyTbYQEBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "RykXd/aXfdqnUkhOUdeFhTgfrHWKtLyUK/AZUO/OSTY="
        },
        "build_page": "/inferences/glm-ocr/base/mradermacher-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/mradermacher/glm-ocr-gguf",
        "weeks": [
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 106.1,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by mradermacher",
        "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.037,
        "cost_breakdown": {
          "tokens_per_sec": 106.1,
          "usd_per_hour": 0.778,
          "usd_per_million": 2.037,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 1.0368
            },
            {
              "resource": "Memory",
              "quantity": "0.5 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.426,
              "usd_per_million": 0.004
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.9425
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0537
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "V/tN/jFjOUBl+pYQsVxuIr+guLrtf7ineA8t0YZRhoiqFy2pfa+VQvCs/oEFBkBygoaEb53lY4MZGjyTbYQEBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "RykXd/aXfdqnUkhOUdeFhTgfrHWKtLyUK/AZUO/OSTY="
        },
        "build_page": "/inferences/glm-ocr/base/mradermacher-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/mradermacher/glm-ocr-gguf",
        "weeks": [
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 2,
      "machines_total": 6,
      "weeks_measured": [
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 106.1,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by mradermacher",
          "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.037,
          "cost_breakdown": {
            "tokens_per_sec": 106.1,
            "usd_per_hour": 0.778,
            "usd_per_million": 2.037,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 1.0368
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.426,
                "usd_per_million": 0.004
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.9425
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0537
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "V/tN/jFjOUBl+pYQsVxuIr+guLrtf7ineA8t0YZRhoiqFy2pfa+VQvCs/oEFBkBygoaEb53lY4MZGjyTbYQEBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "RykXd/aXfdqnUkhOUdeFhTgfrHWKtLyUK/AZUO/OSTY="
          },
          "build_page": "/inferences/glm-ocr/base/mradermacher-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/mradermacher/glm-ocr-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 18.11,
          "context_tokens": 4096,
          "build": "Q3_K_S (3-bit) by mradermacher",
          "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q3_K_S",
          "engine": "ollama",
          "cost_usd_per_million": 3.9367,
          "cost_breakdown": {
            "tokens_per_sec": 18.11,
            "usd_per_hour": 0.2567,
            "usd_per_million": 3.9367,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 3.865
              },
              {
                "resource": "Memory",
                "quantity": "0.4 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.327,
                "usd_per_million": 0.0181
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0537
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "ZsKCFSMr5DcBu/B1ixxi5QwKtsWPLgIp01mije4h3cq/kIsNepOrWOE2ILkEGX1HKAe3I51VpsQ4blRDu8qVBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "pmu2r4qiir7/gUGI9yTh1kVmbCabMNuUgwAehq+ZveE="
          },
          "build_page": "/inferences/glm-ocr/base/mradermacher-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/mradermacher/glm-ocr-gguf",
          "weeks": [
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 13.16,
          "context_tokens": null,
          "build": "Q2_K (2-bit) by mradermacher",
          "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q2_K",
          "engine": "ollama",
          "cost_usd_per_million": 5.4152,
          "cost_breakdown": {
            "tokens_per_sec": 13.16,
            "usd_per_hour": 0.2565,
            "usd_per_million": 5.4152,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 5.32
              },
              {
                "resource": "Memory",
                "quantity": "0.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.28,
                "usd_per_million": 0.0213
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0739
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "uMeGKAvkpkRs1TByJvc65FAWKE+HEk6J9zdLau1lQmVe2iysO5y5JS3QjCiI6ZMS9b2j4EwDMj1eml32dYzyDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "4d+ShCANdBvwSyApV2hlm4WtSFnP8hGhysotsgqECBA="
          },
          "build_page": "/inferences/glm-ocr/base/mradermacher-q2_k-ollama",
          "hub_url": "https://huggingface.co/mradermacher/glm-ocr-gguf",
          "weeks": [
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 12.98,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by mradermacher",
          "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 5.502,
          "cost_breakdown": {
            "tokens_per_sec": 12.98,
            "usd_per_hour": 0.257,
            "usd_per_million": 5.502,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 5.3943
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.426,
                "usd_per_million": 0.0328
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0749
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "1rxnp5E1PvRGeRpVM6osmnrm+ZmNyJ2BUVlmED2i1SbXA1prrFXO9CnIAEsHLnwITXVcc759MDhWmCsVTSblDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "KkcoFM5gxEif40CODU5thjSAebUomuS0+Jsiaxrshrw="
          },
          "build_page": "/inferences/glm-ocr/base/mradermacher-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/mradermacher/glm-ocr-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 4096,
          "build": "W4A16 (4-bit) by AlphaOxO",
          "ref": "AlphaOxO/GLM-OCR-NVFP4",
          "engine": "vllm",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T21:08:21.853Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/glm-ocr/base/alphaoxo-w4a16-vllm",
          "hub_url": "https://huggingface.co/alphaoxo/glm-ocr-nvfp4",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 4096,
          "build": "FP8 (8-bit) by axteam",
          "ref": "axteam/GLM-OCR-FP8-dynamic",
          "engine": "vllm",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T01:45:29.834Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/glm-ocr/base/axteam-fp8-vllm",
          "hub_url": "https://huggingface.co/axteam/glm-ocr-fp8-dynamic",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-OCR"
      },
      "builds": 5,
      "variants": [],
      "bottom_line": "106 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "MedGemma 1.5 4B",
      "made_by": "google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/medgemma-1-5/4b",
      "hub_url": "https://huggingface.co/google/medgemma-1.5-4b-it",
      "download_gb": 2.3,
      "params_b": 4,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 65536,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "reasons about code",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 74.52,
        "context_tokens": 65536,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/medgemma-1.5-4b-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.9204,
        "cost_breakdown": {
          "tokens_per_sec": 74.52,
          "usd_per_hour": 0.7835,
          "usd_per_million": 2.9204,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 1.4761
            },
            {
              "resource": "Memory",
              "quantity": "2.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.932,
              "usd_per_million": 0.0259
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 1.3419
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0764
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "ZsHHtynPKIw3Fmzuq6EyM1UOiq5yUj9LuKWRsrqwyghaUiFGR7CxdoeYeDpYc8e0V7IsrAzhr2gAjH1jSpqYAg==",
          "signing_payload_b64": "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",
          "payload_sha256": "GSsNT9GperqViKgHQMXM9W0aZvaPrFD+wUwOGznNgbQ="
        },
        "build_page": "/inferences/medgemma-1-5/4b/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/medgemma-1.5-4b-it-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 74.52,
        "context_tokens": 65536,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/medgemma-1.5-4b-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.9204,
        "cost_breakdown": {
          "tokens_per_sec": 74.52,
          "usd_per_hour": 0.7835,
          "usd_per_million": 2.9204,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 1.4761
            },
            {
              "resource": "Memory",
              "quantity": "2.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.932,
              "usd_per_million": 0.0259
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 1.3419
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0764
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "ZsHHtynPKIw3Fmzuq6EyM1UOiq5yUj9LuKWRsrqwyghaUiFGR7CxdoeYeDpYc8e0V7IsrAzhr2gAjH1jSpqYAg==",
          "signing_payload_b64": "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",
          "payload_sha256": "GSsNT9GperqViKgHQMXM9W0aZvaPrFD+wUwOGznNgbQ="
        },
        "build_page": "/inferences/medgemma-1-5/4b/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/medgemma-1.5-4b-it-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 74.52,
          "context_tokens": 65536,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/medgemma-1.5-4b-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.9204,
          "cost_breakdown": {
            "tokens_per_sec": 74.52,
            "usd_per_hour": 0.7835,
            "usd_per_million": 2.9204,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 1.4761
              },
              {
                "resource": "Memory",
                "quantity": "2.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.932,
                "usd_per_million": 0.0259
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.3419
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0764
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "ZsHHtynPKIw3Fmzuq6EyM1UOiq5yUj9LuKWRsrqwyghaUiFGR7CxdoeYeDpYc8e0V7IsrAzhr2gAjH1jSpqYAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "GSsNT9GperqViKgHQMXM9W0aZvaPrFD+wUwOGznNgbQ="
          },
          "build_page": "/inferences/medgemma-1-5/4b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/medgemma-1.5-4b-it-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Health Ai Developer Foundations licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/google/medgemma-1.5-4b-it"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "75 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Qwen 3.5 0.8B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3-5/0.8b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.5-0.8B",
      "download_gb": 0.5,
      "params_b": 0.8,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "shows its reasoning",
        "returns valid JSON",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 16.53,
        "context_tokens": null,
        "build": "Q4_0 (4-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q4_0",
        "engine": "ollama-host",
        "cost_usd_per_million": 1.8643,
        "cost_breakdown": {
          "tokens_per_sec": 16.53,
          "usd_per_hour": 0.1109,
          "usd_per_million": 1.8643,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 1.8153
            },
            {
              "resource": "Memory",
              "quantity": "0.5 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.393,
              "usd_per_million": 0.0238
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0252
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "u2R+dnlRZMYS5P/+SLQJPwsSVs7HYAfMXIXJuvX1i3HO7LLuM7qeL+LQt3dareAvXcJQPlCnAoWRNmTAAjOLCg==",
          "signing_payload_b64": "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",
          "payload_sha256": "2szx392CzPrvCGNMacZa2y5IAtUplu2xUq8IwpLzMVs="
        },
        "build_page": "/inferences/qwen-3-5/0.8b/unsloth-q4_0-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3.5-0.8b-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 16.53,
        "context_tokens": null,
        "build": "Q4_0 (4-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q4_0",
        "engine": "ollama-host",
        "cost_usd_per_million": 1.8643,
        "cost_breakdown": {
          "tokens_per_sec": 16.53,
          "usd_per_hour": 0.1109,
          "usd_per_million": 1.8643,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 1.8153
            },
            {
              "resource": "Memory",
              "quantity": "0.5 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.393,
              "usd_per_million": 0.0238
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0252
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "u2R+dnlRZMYS5P/+SLQJPwsSVs7HYAfMXIXJuvX1i3HO7LLuM7qeL+LQt3dareAvXcJQPlCnAoWRNmTAAjOLCg==",
          "signing_payload_b64": "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",
          "payload_sha256": "2szx392CzPrvCGNMacZa2y5IAtUplu2xUq8IwpLzMVs="
        },
        "build_page": "/inferences/qwen-3-5/0.8b/unsloth-q4_0-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3.5-0.8b-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 4,
      "weeks_measured": [
        "2026-W43",
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 16.53,
          "context_tokens": null,
          "build": "Q4_0 (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.8643,
          "cost_breakdown": {
            "tokens_per_sec": 16.53,
            "usd_per_hour": 0.1109,
            "usd_per_million": 1.8643,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.8153
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.393,
                "usd_per_million": 0.0238
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0252
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "u2R+dnlRZMYS5P/+SLQJPwsSVs7HYAfMXIXJuvX1i3HO7LLuM7qeL+LQt3dareAvXcJQPlCnAoWRNmTAAjOLCg==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTYuNTI2NTg4NDE4MDY1NDU4LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5MzQwNTkxNzI5ODY0ODg5MSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMGVlMzVkYzQ2YTIyMTJjYjMxNDQ0Y2E0YzE5NjY5YWM4ZDk3YjA5MjUyMGI4YzllNGQyOWFiNDhkOWVlNjFkM2IiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L25vbmUvMzJnIiwibW9kZWwiOiJoZi5jby91bnNsb3RoL1F3ZW4zLjUtMC44Qi1HR1VGOlE0XzAiLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6MTgsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6MTAxNS45OTkxNjMsInRvdGFsX3Rva2VucyI6MTY3OTF9fQ==",
            "payload_sha256": "2szx392CzPrvCGNMacZa2y5IAtUplu2xUq8IwpLzMVs="
          },
          "build_page": "/inferences/qwen-3-5/0.8b/unsloth-q4_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.5-0.8b-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 2,
          "tokens_per_sec": 16.1,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.9149,
          "cost_breakdown": {
            "tokens_per_sec": 16.1,
            "usd_per_hour": 0.111,
            "usd_per_million": 1.9149,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.8633
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.413,
                "usd_per_million": 0.0257
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0259
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "gpBFJYw0a12RTzr4EKF+YxLNRyADxiuwj3z07LFVxtLOedP8V7DQoKBj/ug55LirhvQkU+pFsWXefqxY3eaCCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "nnvBePZD6Tlgrf/l+r+C7g/kTd59hShv4VjF3yoC740="
          },
          "build_page": "/inferences/qwen-3-5/0.8b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.5-0.8b-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44"
          ],
          "providers": [
            "acbe75",
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 12.25,
          "context_tokens": null,
          "build": "Q8_0 (8-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.5338,
          "cost_breakdown": {
            "tokens_per_sec": 12.25,
            "usd_per_hour": 0.1118,
            "usd_per_million": 2.5338,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.4483
              },
              {
                "resource": "Memory",
                "quantity": "0.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.63,
                "usd_per_million": 0.0514
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.034
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "AaLFtoydR0QYGv+ycnTK3IeSCIt6jZDhKWG1iRBzyVAug6ElXPPAs92XV0f0h07kacGNuDtNAWCcxsb+TXH0DA==",
            "signing_payload_b64": "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",
            "payload_sha256": "uKuG1FEqJOK0rq5bbvISFGdQmKyonqYmGPEFn2cdxCo="
          },
          "build_page": "/inferences/qwen-3-5/0.8b/unsloth-q8_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.5-0.8b-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.5-0.8B"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "17 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "LFM2.5 1.2B",
      "made_by": "LiquidAI",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/61b8e2ba285851687028d395/EsTgVtnM2IqVRKgPdfqcB.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/lfm-2-5/1.2b",
      "hub_url": "https://huggingface.co/LiquidAI/LFM2.5-1.2B-Instruct",
      "download_gb": 0.7,
      "params_b": 1.2,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "reasons about code",
        "returns valid JSON",
        "long answers"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 15.9,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/LFM2.5-1.2B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 1.949,
        "cost_breakdown": {
          "tokens_per_sec": 15.9,
          "usd_per_hour": 0.1115,
          "usd_per_million": 1.949,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 1.8871
            },
            {
              "resource": "Memory",
              "quantity": "0.7 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.567,
              "usd_per_million": 0.0357
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0262
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-25T07:41:14.640Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
          "signature": "o23Daz021BkMs3r0r4bdmCB54kDgtX9N4EJYMyITJ7BqUP96ZRvWIRSWIRW7VbGFjmdHSTsDOQ1oe9Jof8oVBA==",
          "signing_payload_b64": "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",
          "payload_sha256": "fhMb3SBYYz22VO574Jru3iZWyjlOrtcamdB2Xy3v3jk="
        },
        "build_page": "/inferences/lfm-2-5/1.2b/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/lfm2.5-1.2b-instruct-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "acbe75"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 15.9,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/LFM2.5-1.2B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 1.949,
        "cost_breakdown": {
          "tokens_per_sec": 15.9,
          "usd_per_hour": 0.1115,
          "usd_per_million": 1.949,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 1.8871
            },
            {
              "resource": "Memory",
              "quantity": "0.7 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.567,
              "usd_per_million": 0.0357
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0262
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-25T07:41:14.640Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
          "signature": "o23Daz021BkMs3r0r4bdmCB54kDgtX9N4EJYMyITJ7BqUP96ZRvWIRSWIRW7VbGFjmdHSTsDOQ1oe9Jof8oVBA==",
          "signing_payload_b64": "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",
          "payload_sha256": "fhMb3SBYYz22VO574Jru3iZWyjlOrtcamdB2Xy3v3jk="
        },
        "build_page": "/inferences/lfm-2-5/1.2b/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/lfm2.5-1.2b-instruct-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "acbe75"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 15.9,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/LFM2.5-1.2B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.949,
          "cost_breakdown": {
            "tokens_per_sec": 15.9,
            "usd_per_hour": 0.1115,
            "usd_per_million": 1.949,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.8871
              },
              {
                "resource": "Memory",
                "quantity": "0.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.567,
                "usd_per_million": 0.0357
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0262
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "o23Daz021BkMs3r0r4bdmCB54kDgtX9N4EJYMyITJ7BqUP96ZRvWIRSWIRW7VbGFjmdHSTsDOQ1oe9Jof8oVBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "fhMb3SBYYz22VO574Jru3iZWyjlOrtcamdB2Xy3v3jk="
          },
          "build_page": "/inferences/lfm-2-5/1.2b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/lfm2.5-1.2b-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "LFM1.0 licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/LiquidAI/LFM2.5-1.2B-Instruct"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "16 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Magistral Small 2506",
      "made_by": "mistralai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/634c17653d11eaedd88b314d/9OgyfKstSZtbmsmuG8MbU.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/magistral-small/2506",
      "hub_url": "https://huggingface.co/mistralai/Magistral-Small-2506",
      "download_gb": 13.3,
      "params_b": 24,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": 40960,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "returns valid JSON",
        "long answers",
        "summarises sources",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 5.48,
        "context_tokens": 40960,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/Magistral-Small-2506-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 41.3999,
        "cost_breakdown": {
          "tokens_per_sec": 5.48,
          "usd_per_hour": 0.8165,
          "usd_per_million": 41.3999,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 20.0777
            },
            {
              "resource": "Memory",
              "quantity": "13.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 11.124,
              "usd_per_million": 2.0304
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 18.2525
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 1.0393
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "CoalCL8sFFUAgecKAqqZEOc8Ffy1bOMs0FsE8Z9d6BuEzAjIwotV8WoFPi1jeKMexNKqihBx+TKKr+Qci2ozAA==",
          "signing_payload_b64": "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",
          "payload_sha256": "UpvV+HGgEH5SK4FNK981w5hpcktLBxbtSdgiIQRStF8="
        },
        "build_page": "/inferences/magistral-small/2506/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/magistral-small-2506-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 5.48,
        "context_tokens": 40960,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/Magistral-Small-2506-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 41.3999,
        "cost_breakdown": {
          "tokens_per_sec": 5.48,
          "usd_per_hour": 0.8165,
          "usd_per_million": 41.3999,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 20.0777
            },
            {
              "resource": "Memory",
              "quantity": "13.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 11.124,
              "usd_per_million": 2.0304
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 18.2525
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 1.0393
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "CoalCL8sFFUAgecKAqqZEOc8Ffy1bOMs0FsE8Z9d6BuEzAjIwotV8WoFPi1jeKMexNKqihBx+TKKr+Qci2ozAA==",
          "signing_payload_b64": "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",
          "payload_sha256": "UpvV+HGgEH5SK4FNK981w5hpcktLBxbtSdgiIQRStF8="
        },
        "build_page": "/inferences/magistral-small/2506/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/magistral-small-2506-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 5.48,
          "context_tokens": 40960,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Magistral-Small-2506-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 41.3999,
          "cost_breakdown": {
            "tokens_per_sec": 5.48,
            "usd_per_hour": 0.8165,
            "usd_per_million": 41.3999,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 20.0777
              },
              {
                "resource": "Memory",
                "quantity": "13.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 11.124,
                "usd_per_million": 2.0304
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 18.2525
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 1.0393
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "CoalCL8sFFUAgecKAqqZEOc8Ffy1bOMs0FsE8Z9d6BuEzAjIwotV8WoFPi1jeKMexNKqihBx+TKKr+Qci2ozAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "UpvV+HGgEH5SK4FNK981w5hpcktLBxbtSdgiIQRStF8="
          },
          "build_page": "/inferences/magistral-small/2506/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/magistral-small-2506-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/mistralai/Magistral-Small-2506"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "5.5 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Qwen 3.5 35B-A3B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3-5/35b-a3b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.5-35B-A3B",
      "download_gb": 20.5,
      "params_b": 35,
      "active_params_b": 3,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Spanish",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Nepali",
        "Dutch",
        "Punjabi",
        "Sinhala",
        "Tamil",
        "Telugu",
        "Urdu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 3.98,
        "context_tokens": 4096,
        "build": "Q4_K_S (4-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_S",
        "engine": "ollama-host",
        "cost_usd_per_million": 11.6834,
        "cost_breakdown": {
          "tokens_per_sec": 3.98,
          "usd_per_hour": 0.1673,
          "usd_per_million": 11.6834,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 7.5439
            },
            {
              "resource": "Memory",
              "quantity": "19.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 16.045,
              "usd_per_million": 4.0347
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.1049
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD EPYC 7B12",
            "model_raw": "AMD EPYC 7B12",
            "clock_ghz": null,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 512,
            "l3_mb": 16,
            "single_core_hashes_per_sec": 1460.2,
            "multi_core_hashes_per_sec": 3246.1,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T01:38:20.212Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 6.3,
            "latency_ns": 21.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T01:38:21.135Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4982.5,
            "write_mb_s": 179.3,
            "random_read_iops": 476287,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-15T01:38:21.741Z",
            "measured_in": "TW_110",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-15T06:44:44.996Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
          "signature": "ll8T7vTQKj2IOFF+pgq5X8FSdVYG7S7tEpVjkv5MseljkpvIWxK/HCGbTLUC6suVvFWcfKTe+qjRpxcGWXmmAQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "UgzyS5meesDl+TV8ejTVMzsaZH15KN3huzZcsIEWmFc="
        },
        "build_page": "/inferences/qwen-3-5/35b-a3b/unsloth-q4_k_s-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3.5-35b-a3b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "662a4e"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 3.98,
        "context_tokens": 4096,
        "build": "Q4_K_S (4-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_S",
        "engine": "ollama-host",
        "cost_usd_per_million": 11.6834,
        "cost_breakdown": {
          "tokens_per_sec": 3.98,
          "usd_per_hour": 0.1673,
          "usd_per_million": 11.6834,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 7.5439
            },
            {
              "resource": "Memory",
              "quantity": "19.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 16.045,
              "usd_per_million": 4.0347
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.1049
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD EPYC 7B12",
            "model_raw": "AMD EPYC 7B12",
            "clock_ghz": null,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 512,
            "l3_mb": 16,
            "single_core_hashes_per_sec": 1460.2,
            "multi_core_hashes_per_sec": 3246.1,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T01:38:20.212Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 6.3,
            "latency_ns": 21.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T01:38:21.135Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4982.5,
            "write_mb_s": 179.3,
            "random_read_iops": 476287,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-15T01:38:21.741Z",
            "measured_in": "TW_110",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-15T06:44:44.996Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
          "signature": "ll8T7vTQKj2IOFF+pgq5X8FSdVYG7S7tEpVjkv5MseljkpvIWxK/HCGbTLUC6suVvFWcfKTe+qjRpxcGWXmmAQ==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6My45NzY3NDc4MTI3MDIyNjU1LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5NDcyNTA4NDk5Njg5ODc1Niwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDZlZWY2YjczODRiNjY0MjgwZjdhNjIyYjZjMWE4MzFhNDUzMDI5ZDA5ODA5NTZlYzgwOTA5NTRlNzI2NjJhNGUiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L25vbmUvMzJnIiwibW9kZWwiOiJoZi5jby91bnNsb3RoL1F3ZW4zLjUtMzVCLUEzQi1HR1VGOlE0X0tfUyIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50IjoyNywic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjo0MDExLjU2OTQ0MSwidG90YWxfdG9rZW5zIjoxNTk1M319",
          "payload_sha256": "UgzyS5meesDl+TV8ejTVMzsaZH15KN3huzZcsIEWmFc="
        },
        "build_page": "/inferences/qwen-3-5/35b-a3b/unsloth-q4_k_s-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3.5-35b-a3b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "662a4e"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 5,
      "weeks_measured": [
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 3.98,
          "context_tokens": 4096,
          "build": "Q4_K_S (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 11.6834,
          "cost_breakdown": {
            "tokens_per_sec": 3.98,
            "usd_per_hour": 0.1673,
            "usd_per_million": 11.6834,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 7.5439
              },
              {
                "resource": "Memory",
                "quantity": "19.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 16.045,
                "usd_per_million": 4.0347
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1049
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:44:44.996Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "ll8T7vTQKj2IOFF+pgq5X8FSdVYG7S7tEpVjkv5MseljkpvIWxK/HCGbTLUC6suVvFWcfKTe+qjRpxcGWXmmAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "UgzyS5meesDl+TV8ejTVMzsaZH15KN3huzZcsIEWmFc="
          },
          "build_page": "/inferences/qwen-3-5/35b-a3b/unsloth-q4_k_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.5-35b-a3b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "662a4e"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 2,
          "tokens_per_sec": 3.9,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 12.1846,
          "cost_breakdown": {
            "tokens_per_sec": 3.9,
            "usd_per_hour": 0.171,
            "usd_per_million": 12.1846,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 7.6949
              },
              {
                "resource": "Memory",
                "quantity": "20.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 17.087,
                "usd_per_million": 4.3828
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.107
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:44:44.996Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "Du0/LK/kmEofPwPX5IBXRRXpL3ctP6QIaploNM7/Wsh1e+7Z5tTLuXPIjY/y2SOSCeor1+glC/2yGRDN0p7uCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "OB3L4nxoTRZGU3Cspqtrp+iqA55rIvR9fygLBvbt0Ts="
          },
          "build_page": "/inferences/qwen-3-5/35b-a3b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.5-35b-a3b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "662a4e",
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 4096,
          "build": "IQ2_M (2-bit) by unsloth",
          "ref": "unsloth/Qwen3.5-35B-A3B-GGUF:IQ2_M",
          "engine": "llamacpp",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T09:28:40.233Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-3-5/35b-a3b/unsloth-iq2_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.5-35b-a3b-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "IQ2_XXS (2-bit) by unsloth",
          "ref": "unsloth/Qwen3.5-35B-A3B-GGUF:IQ2_XXS",
          "engine": "llamacpp",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T01:52:16.623Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-3-5/35b-a3b/unsloth-iq2_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.5-35b-a3b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.5-35b-a3b",
        "url": "https://openrouter.ai/qwen/qwen3.5-35b-a3b",
        "input_usd_per_million": 0.14,
        "output_usd_per_million": 1,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.065,
            "usd_per_m_out": 0.75,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.112,
            "usd_per_m_out": 0.8,
            "context_length": 262144,
            "uptime_1d": null,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 1,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1,
            "context_length": 256000,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.1625,
            "usd_per_m_out": 1.3,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 1.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 1.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 8
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.5-35B-A3B"
      },
      "builds": 4,
      "variants": [],
      "bottom_line": "4.0 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. About $12 per million tokens here, against OpenRouter's $1.00: the API is 12× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "Qwen3-VL-8B-Instruct-abliterated-v1",
      "made_by": "prithivMLmods",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/65bb837dbfb878f46c77de4c/23gZ_lBEwyoqjexFy9QLD.jpeg",
      "based_on": "Qwen 3 VL-8B",
      "variant_of": "/inferences/qwen-3/vl-8b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/qwen-3/vl-8b~prithivmlmods-abliterated",
      "hub_url": "https://huggingface.co/prithivMLmods/Qwen3-VL-8B-Instruct-abliterated-v1",
      "download_gb": 3.1,
      "params_b": 8,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "returns valid JSON",
        "long answers",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Tamil"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Windows · Intel 1 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel 1 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 1,
          "host": null,
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|1|||"
        },
        "machine_class": "windows|amd64|8|16|intel|1|||",
        "machines": 1,
        "tokens_per_sec": 2.6,
        "context_tokens": 4096,
        "build": "Q2_K (2-bit) by mradermacher",
        "ref": "hf.co/mradermacher/Qwen3-VL-8B-Instruct-abliterated-GGUF:Q2_K",
        "engine": "ollama",
        "cost_usd_per_million": 28.2271,
        "cost_breakdown": {
          "tokens_per_sec": 2.6,
          "usd_per_hour": 0.2647,
          "usd_per_million": 28.2271,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "7 of 8 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 70,
              "usd_per_million": 26.8764
            },
            {
              "resource": "Memory",
              "quantity": "3.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 2.546,
              "usd_per_million": 0.9775
            },
            {
              "resource": "Energy",
              "quantity": "35 W modelled (7 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.972,
              "usd_per_million": 0.3732
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "11th Gen Intel Core i5-1135G7",
            "model_raw": "11th Gen Intel Core i5-1135G7",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 502.2,
            "multi_core_hashes_per_sec": 1190.3,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T12:56:42.595Z",
            "measured_in": "IN_500001",
            "signed": true
          },
          "memory": {
            "total_gb": 15.8,
            "bandwidth_gb_s": 1.2,
            "latency_ns": 253.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T12:56:43.806Z",
            "measured_in": "IN_500001",
            "signed": true
          },
          "storage": {
            "read_mb_s": 329.7,
            "write_mb_s": 12.4,
            "random_read_iops": 8193,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T12:56:49.437Z",
            "measured_in": "IN_500001",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-15T14:04:06.882Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
          "signature": "VLTjDZqtspOKwEDQxQiPMtltKpIBt+S1H4XCnXXpuaoQpmSM2jGsC6T9m7daITKkkQKE3T9r0Sh302ECBXF7AA==",
          "signing_payload_b64": "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",
          "payload_sha256": "Py21e92GPkYTUpiSGvFf4B1o/bqmKCJsUWaM/3SbKls="
        },
        "build_page": "/inferences/qwen-3/vl-8b~prithivmlmods-abliterated/mradermacher-q2_k-ollama",
        "hub_url": "https://huggingface.co/mradermacher/qwen3-vl-8b-instruct-abliterated-gguf",
        "weeks": [
          "2026-W35",
          "2026-W37"
        ],
        "providers": [
          "2e5c02"
        ]
      },
      "cheapest": {
        "machine": "Windows · Intel 1 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel 1 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 1,
          "host": null,
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|1|||"
        },
        "machine_class": "windows|amd64|8|16|intel|1|||",
        "machines": 1,
        "tokens_per_sec": 2.6,
        "context_tokens": 4096,
        "build": "Q2_K (2-bit) by mradermacher",
        "ref": "hf.co/mradermacher/Qwen3-VL-8B-Instruct-abliterated-GGUF:Q2_K",
        "engine": "ollama",
        "cost_usd_per_million": 28.2271,
        "cost_breakdown": {
          "tokens_per_sec": 2.6,
          "usd_per_hour": 0.2647,
          "usd_per_million": 28.2271,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "7 of 8 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 70,
              "usd_per_million": 26.8764
            },
            {
              "resource": "Memory",
              "quantity": "3.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 2.546,
              "usd_per_million": 0.9775
            },
            {
              "resource": "Energy",
              "quantity": "35 W modelled (7 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.972,
              "usd_per_million": 0.3732
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "11th Gen Intel Core i5-1135G7",
            "model_raw": "11th Gen Intel Core i5-1135G7",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 502.2,
            "multi_core_hashes_per_sec": 1190.3,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T12:56:42.595Z",
            "measured_in": "IN_500001",
            "signed": true
          },
          "memory": {
            "total_gb": 15.8,
            "bandwidth_gb_s": 1.2,
            "latency_ns": 253.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T12:56:43.806Z",
            "measured_in": "IN_500001",
            "signed": true
          },
          "storage": {
            "read_mb_s": 329.7,
            "write_mb_s": 12.4,
            "random_read_iops": 8193,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T12:56:49.437Z",
            "measured_in": "IN_500001",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-15T14:04:06.882Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
          "signature": "VLTjDZqtspOKwEDQxQiPMtltKpIBt+S1H4XCnXXpuaoQpmSM2jGsC6T9m7daITKkkQKE3T9r0Sh302ECBXF7AA==",
          "signing_payload_b64": "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",
          "payload_sha256": "Py21e92GPkYTUpiSGvFf4B1o/bqmKCJsUWaM/3SbKls="
        },
        "build_page": "/inferences/qwen-3/vl-8b~prithivmlmods-abliterated/mradermacher-q2_k-ollama",
        "hub_url": "https://huggingface.co/mradermacher/qwen3-vl-8b-instruct-abliterated-gguf",
        "weeks": [
          "2026-W35",
          "2026-W37"
        ],
        "providers": [
          "2e5c02"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W35",
        "2026-W37"
      ],
      "measured_on": [
        {
          "machine": "Windows · Intel 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 1,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|||"
          },
          "machine_class": "windows|amd64|8|16|intel|1|||",
          "machines": 1,
          "tokens_per_sec": 2.6,
          "context_tokens": 4096,
          "build": "Q2_K (2-bit) by mradermacher",
          "ref": "hf.co/mradermacher/Qwen3-VL-8B-Instruct-abliterated-GGUF:Q2_K",
          "engine": "ollama",
          "cost_usd_per_million": 28.2271,
          "cost_breakdown": {
            "tokens_per_sec": 2.6,
            "usd_per_hour": 0.2647,
            "usd_per_million": 28.2271,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 26.8764
              },
              {
                "resource": "Memory",
                "quantity": "3.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 2.546,
                "usd_per_million": 0.9775
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.3732
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "VLTjDZqtspOKwEDQxQiPMtltKpIBt+S1H4XCnXXpuaoQpmSM2jGsC6T9m7daITKkkQKE3T9r0Sh302ECBXF7AA==",
            "signing_payload_b64": "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",
            "payload_sha256": "Py21e92GPkYTUpiSGvFf4B1o/bqmKCJsUWaM/3SbKls="
          },
          "build_page": "/inferences/qwen-3/vl-8b~prithivmlmods-abliterated/mradermacher-q2_k-ollama",
          "hub_url": "https://huggingface.co/mradermacher/qwen3-vl-8b-instruct-abliterated-gguf",
          "weeks": [
            "2026-W35",
            "2026-W37"
          ],
          "providers": [
            "2e5c02"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/prithivMLmods/Qwen3-VL-8B-Instruct-abliterated-v1"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "2.6 tokens a second on Windows · Intel 1 GB · 8 CPUs · 16 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "gemma-4-E2B-it-qat-mobile",
      "made_by": "unsloth",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62ecdc18b72a69615d6bd857/E4lkPz1TZNLzIFr_dR273.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Text & chat",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/unsloth/gemma-4-E2B-it-qat-mobile-GGUF",
      "download_gb": null,
      "params_b": 2,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Malayalam",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "unsloth/gemma-4-E2B-it-qat-mobile-GGUF:Q2_K_XL",
        "engine": "llamacpp",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T03:59:08.665Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/gemma-4/e2b/unsloth-q2_k_xl-ollama",
        "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-qat-mobile-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "unsloth/gemma-4-E2B-it-qat-mobile-GGUF:Q2_K_XL",
          "engine": "llamacpp",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T03:59:08.665Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/gemma-4/e2b/unsloth-q2_k_xl-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-4-e2b-it-qat-mobile-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded."
    },
    {
      "model": "Qwen 3 VL-8B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3/vl-8b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3-VL-8B-Instruct",
      "download_gb": 4.7,
      "params_b": 8,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "returns valid JSON",
        "long answers",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Kannada",
        "Marathi",
        "Punjabi",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": 131072,
        "build": "F16 (16-bit, full precision) by Qwen",
        "ref": "hf.co/Qwen/Qwen3-VL-8B-Instruct-GGUF:F16",
        "engine": "ollama-host",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-23T22:32:04.290Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/qwen-3/vl-8b/qwen-f16-ollama",
        "hub_url": "https://huggingface.co/qwen/qwen3-vl-8b-instruct-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "F16 (16-bit, full precision) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-VL-8B-Instruct-GGUF:F16",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T22:32:04.290Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-3/vl-8b/qwen-f16-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-vl-8b-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3-vl-8b-instruct",
        "url": "https://openrouter.ai/qwen/qwen3-vl-8b-instruct",
        "input_usd_per_million": 0.25,
        "output_usd_per_million": 0.75,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Parasail",
            "tag": "parasail/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.25,
            "usd_per_m_out": 0.75,
            "context_length": 262144,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3-VL-8B-Instruct"
      },
      "builds": 2,
      "variants": [
        {
          "model": "Qwen3-VL-8B-Instruct-abliterated-v1",
          "made_by": "prithivMLmods",
          "page": "/inferences/qwen-3/vl-8b~prithivmlmods-abliterated",
          "status": "measured",
          "tokens_per_sec": 2.6
        }
      ],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded. OpenRouter sells it at $0.75 per million output tokens."
    },
    {
      "model": "FunctionGemma 270M",
      "made_by": "google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/functiongemma/270m",
      "hub_url": "https://huggingface.co/google/functiongemma-270m-it",
      "download_gb": 0.2,
      "params_b": 0.27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 340.82,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.6334,
        "cost_breakdown": {
          "tokens_per_sec": 340.82,
          "usd_per_hour": 0.7772,
          "usd_per_million": 0.6334,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 0.3227
            },
            {
              "resource": "Memory",
              "quantity": "0.2 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.196,
              "usd_per_million": 0.0006
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.2934
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0167
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "VCdNM5VrOGUuczk4NN28xP8Nl9jB+1kfCsrtkodSKIUketpEGf2CEpEV/i6bXq4lzOPVswFvfzsM3er1jegYBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "W3nVeObvEQ2mzicETEDWcLKx7EWwJKsK5qdvjaJRh9k="
        },
        "build_page": "/inferences/functiongemma/270m/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/functiongemma-270m-it-gguf",
        "weeks": [
          "2026-W43",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 340.82,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.6334,
        "cost_breakdown": {
          "tokens_per_sec": 340.82,
          "usd_per_hour": 0.7772,
          "usd_per_million": 0.6334,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 0.3227
            },
            {
              "resource": "Memory",
              "quantity": "0.2 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.196,
              "usd_per_million": 0.0006
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.2934
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0167
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "VCdNM5VrOGUuczk4NN28xP8Nl9jB+1kfCsrtkodSKIUketpEGf2CEpEV/i6bXq4lzOPVswFvfzsM3er1jegYBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "W3nVeObvEQ2mzicETEDWcLKx7EWwJKsK5qdvjaJRh9k="
        },
        "build_page": "/inferences/functiongemma/270m/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/functiongemma-270m-it-gguf",
        "weeks": [
          "2026-W43",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 3,
      "machines_total": 16,
      "weeks_measured": [
        "2026-W43",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 340.82,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.6334,
          "cost_breakdown": {
            "tokens_per_sec": 340.82,
            "usd_per_hour": 0.7772,
            "usd_per_million": 0.6334,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 0.3227
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.0006
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.2934
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0167
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "VCdNM5VrOGUuczk4NN28xP8Nl9jB+1kfCsrtkodSKIUketpEGf2CEpEV/i6bXq4lzOPVswFvfzsM3er1jegYBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "W3nVeObvEQ2mzicETEDWcLKx7EWwJKsK5qdvjaJRh9k="
          },
          "build_page": "/inferences/functiongemma/270m/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 66.76,
          "context_tokens": null,
          "build": "IQ2_XXS (2-bit) by unsloth",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:IQ2_XXS",
          "engine": "ollama",
          "cost_usd_per_million": 1.0652,
          "cost_breakdown": {
            "tokens_per_sec": 66.76,
            "usd_per_hour": 0.256,
            "usd_per_million": 1.0652,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 1.0485
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.139,
                "usd_per_million": 0.0021
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0146
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "aqDiwuUniKURoR8YnkaDL6q78AStiinzUVA0ABckw67aFt7NwzaJID6LC0+46RQEy8DUe8fMGxChQXqMprolDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "d+rby3WMXs+DyBdUeVYzEVrw/+yzeGSjcAu+8wqaY20="
          },
          "build_page": "/inferences/functiongemma/270m/unsloth-iq2_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 40.24,
          "context_tokens": null,
          "build": "IQ4_NL (4-bit) by Mungert",
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:IQ4_NL",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.7588,
          "cost_breakdown": {
            "tokens_per_sec": 40.24,
            "usd_per_hour": 0.1099,
            "usd_per_million": 0.7588,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.7455
              },
              {
                "resource": "Memory",
                "quantity": "0.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.122,
                "usd_per_million": 0.003
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0104
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "12rypP10l3lR7QSQU36TUIklHhKvho4HxHulpCLRay2ojtKf2C1dfuOS6EHfXpEnCRWcMcJ/9ZplRIrfn20tAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "1ZdtjM17nqHoFZkclvLkt5mdyx8tH8OKJBcTHttsLco="
          },
          "build_page": "/inferences/functiongemma/270m~mungert/mungert-iq4_nl-ollama",
          "hub_url": "https://huggingface.co/mungert/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 37.25,
          "context_tokens": null,
          "build": "Q4_0 (4-bit) by unsloth",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.8216,
          "cost_breakdown": {
            "tokens_per_sec": 37.25,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.8216,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8054
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.187,
                "usd_per_million": 0.005
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0112
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "1Xr6ZK4G15aTY71zCNrHq0IwNDP7RglJc7tsOZK3S7HbWLN/JiPYrMxvWoK8hrOEYacRd1xU71H5qbxZhzn9CA==",
            "signing_payload_b64": "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",
            "payload_sha256": "GDlirxBJhckrZHD4rJNMTAp+z+GURfIutfWxrW8z3VA="
          },
          "build_page": "/inferences/functiongemma/270m/unsloth-q4_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 36.7,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by victor",
          "ref": "hf.co/victor/functiongemma-agent-gguf:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.834,
          "cost_breakdown": {
            "tokens_per_sec": 36.7,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.834,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8173
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.0053
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0114
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "sdK4piV7QQ+QfjlKJLTTNukUsM+uXtRAeBzz8nPfLbd9icdJM8FkdTDIs8dlyCIt+JQF13PLxNH6nC/F6JCHCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "zjwMzhVd0JAHWa40xqRET+b9qUJ40PyJUopzdPcItVg="
          },
          "build_page": "/inferences/functiongemma/270m~unsloth/victor-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/victor/functiongemma-agent-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 34.43,
          "context_tokens": null,
          "build": "Q3_K_S (3-bit) by Mungert",
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.8878,
          "cost_breakdown": {
            "tokens_per_sec": 34.43,
            "usd_per_hour": 0.11,
            "usd_per_million": 0.8878,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8714
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.146,
                "usd_per_million": 0.0042
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0121
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "9qMQqd9D6z6uOwkHDEhTOWNv+FEOlBQriHP6KlSGeXTEi8em0grsT9GwiUvOhmk40y3IzyL9bYxPxSuQrbYeDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "XrhOQ9yDiiOns7V/FIMu3gyylMs/+tFAgWxq3RzJx2o="
          },
          "build_page": "/inferences/functiongemma/270m~mungert/mungert-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/mungert/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 33.83,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9049,
          "cost_breakdown": {
            "tokens_per_sec": 33.83,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.9049,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8868
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.0058
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0123
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "TqjB1okKtMfuq9sZOu4z9S/AVZ1cgwWGNnzkKtjTjEEeUk6wdGI2m7oTB+XkNSo3YZ02Pk3IMy28aLsCkshKDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "wdVXt8Aij+TYoK0GNFncL4QqU47ZqUrySoXnvNOkTME="
          },
          "build_page": "/inferences/functiongemma/270m/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 33.51,
          "context_tokens": null,
          "build": "Q4_1 (4-bit) by Mungert",
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q4_1",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.912,
          "cost_breakdown": {
            "tokens_per_sec": 33.51,
            "usd_per_hour": 0.11,
            "usd_per_million": 0.912,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8953
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.143,
                "usd_per_million": 0.0043
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0124
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "pcJRXb0qh1NagvQv1GdgbZeSwy4KIW3cROIGdQp7uEmGjnI6Yfxkzfki2MQ9YvxXwLGpMHLP7tRjuyUcEVoqBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "728EVkpbzQsIck8f4FjVulAsSiR5ve/AndwqpCU+Adc="
          },
          "build_page": "/inferences/functiongemma/270m~mungert/mungert-q4_1-ollama",
          "hub_url": "https://huggingface.co/mungert/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 32.9,
          "context_tokens": null,
          "build": "Q8_0 (8-bit) by unsloth",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9315,
          "cost_breakdown": {
            "tokens_per_sec": 32.9,
            "usd_per_hour": 0.1103,
            "usd_per_million": 0.9315,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.912
              },
              {
                "resource": "Memory",
                "quantity": "0.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.226,
                "usd_per_million": 0.0069
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0127
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "Tea2B/IZ9A5RGXeqG14BFAEjri4XAAV4lBCG3jrBmoWwtDwhIgBnlrb/hj6jHa2XmCfYD7gQPCo/B/+LBU/VBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "8GJYxGqZkmPjaTv3a7vX/jtRzNKE7/V1/JHeN7LDrfI="
          },
          "build_page": "/inferences/functiongemma/270m/unsloth-q8_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 32.75,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by Mungert",
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9347,
          "cost_breakdown": {
            "tokens_per_sec": 32.75,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.9347,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.916
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.006
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0127
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "jleFQ4bmkTut9BEqG024uX94IWTNbD2w3gOZ0tnkg/cuUtjA3+TiBWPVBLir5500ExVDd7/SWpC3TZw9NpCIAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "/hM3vUAwHMEfRGl2otrZOWOQxYQ648wR68cDrWKR9uE="
          },
          "build_page": "/inferences/functiongemma/270m~mungert/mungert-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/mungert/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 32.63,
          "context_tokens": null,
          "build": "IQ3_XXS (3-bit) by unsloth",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9367,
          "cost_breakdown": {
            "tokens_per_sec": 32.63,
            "usd_per_hour": 0.11,
            "usd_per_million": 0.9367,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.9195
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.142,
                "usd_per_million": 0.0044
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0128
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "d4CMG9d0or8ou2f0tEr4nJABR0q31ojLYldwMA60ADcrMyE4nPuA8EFb2kIdye3NLJ6Tz+LoyLC9I2Ir9IfKCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "3IzfEQcWPe+PgogAfbl4At74MqumxY86mYuFOntJWhw="
          },
          "build_page": "/inferences/functiongemma/270m/unsloth-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 32.5,
          "context_tokens": null,
          "build": "IQ2_M (2-bit) by unsloth",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:IQ2_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9401,
          "cost_breakdown": {
            "tokens_per_sec": 32.5,
            "usd_per_hour": 0.11,
            "usd_per_million": 0.9401,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.923
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.141,
                "usd_per_million": 0.0043
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0128
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "LltJZgqjdcAsDLFka4VTfaXG+H2oX78veE5b6ekNbj4UnzXmfD55xgLOj8bXAzRsDQHNdqFlgWtI3mv7CbYvCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "Ip0Y7iDisyedZcIBG1ug30CZ1pJjM744QT3LSDissJ4="
          },
          "build_page": "/inferences/functiongemma/270m/unsloth-iq2_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 23.66,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 3.0082,
          "cost_breakdown": {
            "tokens_per_sec": 23.66,
            "usd_per_hour": 0.2562,
            "usd_per_million": 3.0082,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 2.9588
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.0083
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0411
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "jvSCo5FLY3tF4kRI/BluRZV2pZnl1EqCYwXzHoEr4w5dv4vVnm51L1E4ut8+mqNu2GJ8y0bPim5/h0v/FlNNAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "ehDr29HCZkLDu8m0JS7mbVw6XxFvdLSVb5p/QA7uHB8="
          },
          "build_page": "/inferences/functiongemma/270m/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 22.34,
          "context_tokens": null,
          "build": "Q8_0 (8-bit) by Mungert",
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.3783,
          "cost_breakdown": {
            "tokens_per_sec": 22.34,
            "usd_per_hour": 0.1109,
            "usd_per_million": 1.3783,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.3428
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.376,
                "usd_per_million": 0.0168
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0187
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "I0/JT4uWZGByqIiKOyCbttKuIDIrpwY4n5zL6rbtZ7CapLZSEqB0rinhcMqRraRQmbWzTmNl2YXz/BBLs5A2DQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "2xLRKMAk2QyFmDNw8bMAmSXivnd8DQOsMua+TxzxHQY="
          },
          "build_page": "/inferences/functiongemma/270m~mungert/mungert-q8_0-ollama",
          "hub_url": "https://huggingface.co/mungert/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 20.08,
          "context_tokens": null,
          "build": "BF16 (16-bit, full precision) by Mungert",
          "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:BF16",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.5355,
          "cost_breakdown": {
            "tokens_per_sec": 20.08,
            "usd_per_hour": 0.111,
            "usd_per_million": 1.5355,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.4937
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.421,
                "usd_per_million": 0.021
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0208
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "RI2HqiJ/k+JkKZrbHt+J7s+cFmkz1fpXQ4b1VovtIN5k73Bs9Myi4y54qKwUybUJkKPbVSQHW8GAaca+H9GuDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "/EIt/v1FEtqR6Q9k8ui4/gcCs0ER6IsWkVEF+ZbLLfs="
          },
          "build_page": "/inferences/functiongemma/270m~mungert/mungert-bf16-ollama",
          "hub_url": "https://huggingface.co/mungert/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 18.71,
          "context_tokens": null,
          "build": "BF16 (16-bit, full precision) by unsloth",
          "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:BF16",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.6483,
          "cost_breakdown": {
            "tokens_per_sec": 18.71,
            "usd_per_hour": 0.111,
            "usd_per_million": 1.6483,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.6035
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.421,
                "usd_per_million": 0.0225
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0223
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "FUseo8mhzweHoLuZC97cBuMNEATK8ae5ouOqaYniRJWqL9VJbByDhRW+gKWR+MUelLN8lfWv/e2zuuoc5Ft1Dw==",
            "signing_payload_b64": "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",
            "payload_sha256": "hRd/ZmBQTafxmCDri2ZDC9h+DRATq3Nddu1d4mThe8U="
          },
          "build_page": "/inferences/functiongemma/270m/unsloth-bf16-ollama",
          "hub_url": "https://huggingface.co/unsloth/functiongemma-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Gemma terms of use",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/Mungert/functiongemma-270m-it-GGUF"
      },
      "builds": 14,
      "variants": [
        {
          "model": "functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
          "made_by": "mingxilei",
          "page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat",
          "status": "measured",
          "tokens_per_sec": 39.3
        },
        {
          "model": "Home-FunctionGemma-270m",
          "made_by": "acon96",
          "page": "/inferences/functiongemma/270m~acon96-home",
          "status": "measured",
          "tokens_per_sec": 35.88
        },
        {
          "model": "functiongemma-270m-it-abominable-heresy",
          "made_by": "MuXodious",
          "page": "/inferences/functiongemma/270m~muxodious-abominable-heresy",
          "status": "measured",
          "tokens_per_sec": 38.58
        }
      ],
      "bottom_line": "341 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Phi 4 Mini-Reasoning",
      "made_by": "microsoft",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1583646260758-5e64858c87403103f9f1055d.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/phi-4/mini-reasoning",
      "hub_url": "https://huggingface.co/microsoft/Phi-4-mini-reasoning",
      "download_gb": 2.3,
      "params_b": 4.1,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "reasons about code",
        "shows its reasoning",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "English",
        "Tamil",
        "Spanish"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 12,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|||"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|||",
        "machines": 1,
        "tokens_per_sec": 189.43,
        "context_tokens": 4096,
        "build": "IQ3_XXS (3-bit) by unsloth",
        "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
        "engine": "ollama",
        "cost_usd_per_million": 1.3595,
        "cost_breakdown": {
          "tokens_per_sec": 189.43,
          "usd_per_hour": 0.9272,
          "usd_per_million": 1.3595,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "15 of 16 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 150,
              "usd_per_million": 0.7918
            },
            {
              "resource": "Memory",
              "quantity": "1.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.295,
              "usd_per_million": 0.0068
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.5279
            },
            {
              "resource": "Energy",
              "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 6.25,
              "usd_per_million": 0.033
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-10T11:20:52.995Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "kd3dBY+d0Z9SiYQZuI1Vs4Zn7iaYvzy+zc/iMfSpNu2J0mmXuOXhNqTLyks0ZNqD2jb1WmXj1IcSHOM5g7JhDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "hiu1HBlqSAgtbPIV5+zmJPT/zyjyiVdlXE6V77H83B0="
        },
        "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq3_xxs-ollama",
        "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
        "weeks": [
          "2026-W37",
          "2026-W38",
          "2026-W41",
          "2026-W42"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "cheapest": {
        "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 12,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|||"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|||",
        "machines": 1,
        "tokens_per_sec": 189.43,
        "context_tokens": 4096,
        "build": "IQ3_XXS (3-bit) by unsloth",
        "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
        "engine": "ollama",
        "cost_usd_per_million": 1.3595,
        "cost_breakdown": {
          "tokens_per_sec": 189.43,
          "usd_per_hour": 0.9272,
          "usd_per_million": 1.3595,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "15 of 16 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 150,
              "usd_per_million": 0.7918
            },
            {
              "resource": "Memory",
              "quantity": "1.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.295,
              "usd_per_million": 0.0068
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.5279
            },
            {
              "resource": "Energy",
              "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 6.25,
              "usd_per_million": 0.033
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-10T11:20:52.995Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "kd3dBY+d0Z9SiYQZuI1Vs4Zn7iaYvzy+zc/iMfSpNu2J0mmXuOXhNqTLyks0ZNqD2jb1WmXj1IcSHOM5g7JhDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "hiu1HBlqSAgtbPIV5+zmJPT/zyjyiVdlXE6V77H83B0="
        },
        "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq3_xxs-ollama",
        "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
        "weeks": [
          "2026-W37",
          "2026-W38",
          "2026-W41",
          "2026-W42"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "machine_kinds": 9,
      "machines_total": 23,
      "weeks_measured": [
        "2026-W34",
        "2026-W35",
        "2026-W37",
        "2026-W38",
        "2026-W40",
        "2026-W41",
        "2026-W42"
      ],
      "measured_on": [
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 189.43,
          "context_tokens": 4096,
          "build": "IQ3_XXS (3-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
          "engine": "ollama",
          "cost_usd_per_million": 1.3595,
          "cost_breakdown": {
            "tokens_per_sec": 189.43,
            "usd_per_hour": 0.9272,
            "usd_per_million": 1.3595,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.7918
              },
              {
                "resource": "Memory",
                "quantity": "1.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.295,
                "usd_per_million": 0.0068
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.5279
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.033
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "kd3dBY+d0Z9SiYQZuI1Vs4Zn7iaYvzy+zc/iMfSpNu2J0mmXuOXhNqTLyks0ZNqD2jb1WmXj1IcSHOM5g7JhDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "hiu1HBlqSAgtbPIV5+zmJPT/zyjyiVdlXE6V77H83B0="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W41",
            "2026-W42"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 89.92,
          "context_tokens": null,
          "build": "Q3_K_L (3-bit) by lmstudio-community",
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q3_K_L",
          "engine": "ollama",
          "cost_usd_per_million": 2.8497,
          "cost_breakdown": {
            "tokens_per_sec": 89.92,
            "usd_per_hour": 0.9225,
            "usd_per_million": 2.8497,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 1.6681
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.1121
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0695
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "MNjOUJ1l6SEBe4q5ADBGa1k3Mes7L+Mc4zpyIXCEtRp8wPvYrGqc0LaTh/ep7kJ1yhwJbxi1DXxmWQuxttRmDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "+g7uZBMLADV28v3ebKfzsovu00t7yOky2RSlWr+Bjh8="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/lmstudio-community-q3_k_l-ollama",
          "hub_url": "https://huggingface.co/lmstudio-community/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W41"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · NVIDIA 4 GB · 12 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 4 GB · 12 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 12,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 4,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|12|16|nvidia|4|||"
          },
          "machine_class": "windows|amd64|12|16|nvidia|4|||",
          "machines": 1,
          "tokens_per_sec": 41.79,
          "context_tokens": 4096,
          "build": "IQ3_XXS (3-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "cost_usd_per_million": 5.1926,
          "cost_breakdown": {
            "tokens_per_sec": 41.79,
            "usd_per_hour": 0.7812,
            "usd_per_million": 5.1926,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 2.6323
              },
              {
                "resource": "Memory",
                "quantity": "1.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.295,
                "usd_per_million": 0.031
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.393
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.1363
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i7-11600H",
              "model_raw": "11th Gen Intel Core i7-11600H",
              "clock_ghz": null,
              "cores_logical": 12,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 365.3,
              "multi_core_hashes_per_sec": 3482.5,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-01T21:15:49.919Z",
              "measured_in": "IN_560037",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.2,
              "latency_ns": 44.2,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-01T21:15:50.572Z",
              "measured_in": "IN_560037",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3050 Laptop GPU",
              "memory_gb": 4,
              "fp16_tflops": 0,
              "watts": null,
              "tflops_per_watt": null,
              "method": null,
              "measured_at": "2026-09-11T19:24:20.219Z",
              "measured_in": null,
              "signed": true
            },
            "storage": {
              "read_mb_s": 1409.8,
              "write_mb_s": 340.5,
              "random_read_iops": 46320,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-01T21:15:51.083Z",
              "measured_in": "IN_560037",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-01T21:35:20.666Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201844bc3835f475fd8afa64d04ddcb9a1e0087bb40a31973d91b193e6c953b229",
            "signature": "gDRvtw7fPU4gl1dM/zjUVcGGX2ETeMmCXTafGYhS0gvdC+3fqBeilyANhBZ7ZHWl8xbPqPNSAW1iXDl/JmamBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "WpmA3Ye/biYhvXQ85EuZ+5tdogk9mndQsW5Sm2HecZ4="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W41",
            "2026-W42"
          ],
          "providers": [
            "53b229"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 25.64,
          "context_tokens": 4096,
          "build": "IQ3_XXS (3-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "cost_usd_per_million": 7.6718,
          "cost_breakdown": {
            "tokens_per_sec": 25.64,
            "usd_per_hour": 0.7082,
            "usd_per_million": 7.6718,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 3.51
              },
              {
                "resource": "Memory",
                "quantity": "1.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.295,
                "usd_per_million": 0.0505
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 3.9
              },
              {
                "resource": "Energy",
                "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.417,
                "usd_per_million": 0.2113
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "IBMT/F7hJO2nqME2KrP/n7yY+nlYuLWtpwc6ibnWXclQaSoL72bjvCVgWe9DcS8bcwOiIowpAdTCxyxLS9/KBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "zw7JoC+W1ts5fP31Khuov5RNyFYYQkMnRNGNrRlI+r0="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W41",
            "2026-W42"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 22.03,
          "context_tokens": 4096,
          "build": "Q8_0 (8-bit) by lmstudio-community",
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 8.8724,
          "cost_breakdown": {
            "tokens_per_sec": 22.03,
            "usd_per_hour": 0.7035,
            "usd_per_million": 8.8724,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 4.0862
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 4.5402
              },
              {
                "resource": "Energy",
                "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.417,
                "usd_per_million": 0.2459
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "oLcOs1mva6Fmz/UyzJDs93C0Rw0fLUF8MmxM9JejJ5dsPwzr4DxwOuIkJuFh1NpA+RyYeCxXP0bXSyIirZ9EBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "tqbJL0BnbZueOPwGU0LF5uuM6hhffsKnwoKH4BPKSko="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/lmstudio-community-q8_0-ollama",
          "hub_url": "https://huggingface.co/lmstudio-community/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W40"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 17.18,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ1_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 11.4266,
          "cost_breakdown": {
            "tokens_per_sec": 17.18,
            "usd_per_hour": 0.7067,
            "usd_per_million": 11.4266,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 5.2388
              },
              {
                "resource": "Memory",
                "quantity": "1.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.885,
                "usd_per_million": 0.0515
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 5.8209
              },
              {
                "resource": "Energy",
                "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.417,
                "usd_per_million": 0.3153
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "TgIMAsbsjGLzqcoXlmZgfCLAc6/lk525Hj4VyOI1LwtKD8f67ylHa9TBfFohziMxZGK4oFb70A8iKZoFzt89Aw==",
            "signing_payload_b64": "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",
            "payload_sha256": "38Ohfy5RfNrF/bCYGGoJbq3IwacEhCtW8VQE2rYD/QE="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq1_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W37",
            "2026-W38",
            "2026-W41",
            "2026-W42"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Windows · NVIDIA 4 GB · 12 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 4 GB · 12 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 12,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 4,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|12|16|nvidia|4|||"
          },
          "machine_class": "windows|amd64|12|16|nvidia|4|||",
          "machines": 1,
          "tokens_per_sec": 16.34,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ1_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 13.2553,
          "cost_breakdown": {
            "tokens_per_sec": 16.34,
            "usd_per_hour": 0.7797,
            "usd_per_million": 13.2553,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 6.7323
              },
              {
                "resource": "Memory",
                "quantity": "1.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.885,
                "usd_per_million": 0.0542
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 6.1203
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.3485
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i7-11600H",
              "model_raw": "11th Gen Intel Core i7-11600H",
              "clock_ghz": null,
              "cores_logical": 12,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 365.3,
              "multi_core_hashes_per_sec": 3482.5,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-01T21:15:49.919Z",
              "measured_in": "IN_560037",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.2,
              "latency_ns": 44.2,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-01T21:15:50.572Z",
              "measured_in": "IN_560037",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3050 Laptop GPU",
              "memory_gb": 4,
              "fp16_tflops": 0,
              "watts": null,
              "tflops_per_watt": null,
              "method": null,
              "measured_at": "2026-09-11T19:24:20.219Z",
              "measured_in": null,
              "signed": true
            },
            "storage": {
              "read_mb_s": 1409.8,
              "write_mb_s": 340.5,
              "random_read_iops": 46320,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-01T21:15:51.083Z",
              "measured_in": "IN_560037",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-01T21:35:20.666Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201844bc3835f475fd8afa64d04ddcb9a1e0087bb40a31973d91b193e6c953b229",
            "signature": "zCysT6iezXgDmBvHiRfSiVeRLj0yHgn6apQPaWwFrUaEurTZUetY19HtFKp0ZjGPunf2THurbziRV19oS8H8BQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "t8mOB8pC2oyq6NIBQ7aeeHLDY2QGQregUa2R22liNGY="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq1_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W37",
            "2026-W38",
            "2026-W41",
            "2026-W42"
          ],
          "providers": [
            "53b229"
          ]
        },
        {
          "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "windows|amd64|8|16|none|0|||"
          },
          "machine_class": "windows|amd64|8|16|none|0|||",
          "machines": 1,
          "tokens_per_sec": 9.83,
          "context_tokens": null,
          "build": "Q2_K (2-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q2_K",
          "engine": "ollama",
          "cost_usd_per_million": 7.3078,
          "cost_breakdown": {
            "tokens_per_sec": 9.83,
            "usd_per_hour": 0.2587,
            "usd_per_million": 7.3078,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 7.1189
              },
              {
                "resource": "Memory",
                "quantity": "1.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.885,
                "usd_per_million": 0.09
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0989
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "b/w+k4PL1xv5wjvHmhr8aXL+R/9fN/y4D7vS4+w0plH5EcOXRSkVPQ+X1uPkAixob3TQaRKgUOCLcwNgNpwcDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "CeE2caDeuSRVRHmWnmkta8KGkKBOt9jqSe6lkM4Skvo="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-q2_k-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W35"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 8 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 8 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 15.7.3",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|8|8|apple|0|||"
          },
          "machine_class": "darwin|arm64|8|8|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 7.26,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by lmstudio-community",
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 24.1396,
          "cost_breakdown": {
            "tokens_per_sec": 7.26,
            "usd_per_hour": 0.6305,
            "usd_per_million": 24.1396,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 9.6482
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 13.7831
              },
              {
                "resource": "Energy",
                "quantity": "185 W modelled (7 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.139,
                "usd_per_million": 0.7083
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-08T15:04:04.375Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ab92819c6391e87d553e27cec9a18a52937a84a123bf752d00c2aa2bee0548a8",
            "signature": "gx1LEU02dfkRtRmuI0d2Aawbv054q5xHhItllU2vE6Qh3XqovnbFFbhKsvOdgF9yRgrezd3T4sAcZi7I+BKfCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "O3xHJzTRlWoGTgNmPfzlwPAfucXKmYvQg7+7apV4iiU="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/lmstudio-community-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/lmstudio-community/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W38",
            "2026-W41"
          ],
          "providers": [
            "0548a8"
          ]
        },
        {
          "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "windows|amd64|8|16|none|0|||"
          },
          "machine_class": "windows|amd64|8|16|none|0|||",
          "machines": 1,
          "tokens_per_sec": 6.5,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 11.221,
          "cost_breakdown": {
            "tokens_per_sec": 6.5,
            "usd_per_hour": 0.2625,
            "usd_per_million": 11.221,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 10.7737
              },
              {
                "resource": "Memory",
                "quantity": "2.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.934,
                "usd_per_million": 0.2977
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.1496
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "WZAvN6ZkpZQC4GFjGuWfJfOE4iB7/kpD/hvNHu1bPjwOd6U3N4NAwNMbgHrW9mMBNQJ/vjHlP7JzWYR7OrdqAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "W/IEWu7EX7sLeyrVp1DbaZYSgJUfhLyVtlcPjRTahoo="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W35"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": 5.6,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ1_S",
          "engine": "ollama",
          "cost_usd_per_million": 12.8228,
          "cost_breakdown": {
            "tokens_per_sec": 5.6,
            "usd_per_hour": 0.2587,
            "usd_per_million": 12.8228,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 12.4914
              },
              {
                "resource": "Memory",
                "quantity": "1.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.885,
                "usd_per_million": 0.1579
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.1735
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "8fcnRaQ3m3MA0c8FUGdMkij/UPytVibwj2cQdk1FKhBz7CDPbcnWm2se6wyNxw+nyoYYQO1NYS++pg7w8TYLCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "vmXoshOOe7Fs0NKATmXey2PaLZee+Nw0RYgUwyEVb+Y="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq1_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W37",
            "2026-W38",
            "2026-W41",
            "2026-W42"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Windows · AMD 1 GB · 8 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · AMD 1 GB · 8 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.22631",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 7,
            "gpu_vendor": "AMD",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 1,
            "host": null,
            "vendor_org": "amd",
            "host_org": null,
            "key": "windows|amd64|8|7|amd|1|||"
          },
          "machine_class": "windows|amd64|8|7|amd|1|||",
          "machines": 1,
          "tokens_per_sec": 4.75,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by lmstudio-community",
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 36.8345,
          "cost_breakdown": {
            "tokens_per_sec": 4.75,
            "usd_per_hour": 0.6305,
            "usd_per_million": 36.8345,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 14.7221
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 21.0316
              },
              {
                "resource": "Energy",
                "quantity": "185 W modelled (7 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.139,
                "usd_per_million": 1.0808
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 997.6,
              "multi_core_hashes_per_sec": 8516.9,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T08:14:32.559Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "memory": {
              "total_gb": 7,
              "bandwidth_gb_s": 6,
              "latency_ns": 61.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T08:14:33.251Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2347.3,
              "write_mb_s": 204.2,
              "random_read_iops": 96070,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-10-29T08:14:33.869Z",
              "measured_in": "IN_560001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-29T09:04:03.140Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
            "signature": "bz55s8iLtRwoJxOB9SdY7afyef7DIgX/JsCEU1fG2S1Ql+H6Knbf79FmxURFX8A1kebBXMRW5wZQURTa5etqDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "5GSE+TKMXES9NNa4WCvTNhc6p7CNe942CMIV3ETrgCk="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/lmstudio-community-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/lmstudio-community/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W38",
            "2026-W41"
          ],
          "providers": [
            "d061c9"
          ]
        },
        {
          "machine": "Windows · AMD 1 GB · 8 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · AMD 1 GB · 8 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.22631",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 7,
            "gpu_vendor": "AMD",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 1,
            "host": null,
            "vendor_org": "amd",
            "host_org": null,
            "key": "windows|amd64|8|7|amd|1|||"
          },
          "machine_class": "windows|amd64|8|7|amd|1|||",
          "machines": 1,
          "tokens_per_sec": 3.85,
          "context_tokens": null,
          "build": "Q3_K_L (3-bit) by lmstudio-community",
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "cost_usd_per_million": 45.5148,
          "cost_breakdown": {
            "tokens_per_sec": 3.85,
            "usd_per_hour": 0.6305,
            "usd_per_million": 45.5148,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 18.1915
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 25.9878
              },
              {
                "resource": "Energy",
                "quantity": "185 W modelled (7 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.139,
                "usd_per_million": 1.3355
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 997.6,
              "multi_core_hashes_per_sec": 8516.9,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T08:14:32.559Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "memory": {
              "total_gb": 7,
              "bandwidth_gb_s": 6,
              "latency_ns": 61.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T08:14:33.251Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2347.3,
              "write_mb_s": 204.2,
              "random_read_iops": 96070,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-10-29T08:14:33.869Z",
              "measured_in": "IN_560001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-29T09:04:03.140Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
            "signature": "HS/TRAZZPq/fkBoHbvUMR9LshMP4xyK3ljVAKV9RW9BWk/XkMlDOYLgUCxTFMI/WoRe19yY8lFDp49J48dN+Dg==",
            "signing_payload_b64": "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",
            "payload_sha256": "pVDxiu9t7Zf3V7Fn4FnHMAn0EixHjlaeEcGoLzk3+Ng="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/lmstudio-community-q3_k_l-ollama",
          "hub_url": "https://huggingface.co/lmstudio-community/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W41"
          ],
          "providers": [
            "d061c9"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 8 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 8 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 15.7.3",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|8|8|apple|0|||"
          },
          "machine_class": "darwin|arm64|8|8|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 3.42,
          "context_tokens": null,
          "build": "Q3_K_L (3-bit) by lmstudio-community",
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "cost_usd_per_million": 51.1518,
          "cost_breakdown": {
            "tokens_per_sec": 3.42,
            "usd_per_hour": 0.6305,
            "usd_per_million": 51.1518,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 20.4445
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 29.2064
              },
              {
                "resource": "Energy",
                "quantity": "185 W modelled (7 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.139,
                "usd_per_million": 1.5009
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-08T15:04:04.375Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ab92819c6391e87d553e27cec9a18a52937a84a123bf752d00c2aa2bee0548a8",
            "signature": "jkZhZ1DlFxGRDHhUvtlbWAUYVA4gk5qsmuAVs2J2zHAs0M6+ccw/yJsEagHYtdajmpg180ccgTs0jG1MHJhaBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "93PcLOh/htMZt5b+qnibG7MDO0JlUYRgvS6tN00/h6E="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/lmstudio-community-q3_k_l-ollama",
          "hub_url": "https://huggingface.co/lmstudio-community/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W41"
          ],
          "providers": [
            "0548a8"
          ]
        },
        {
          "machine": "Mac · no GPU · 4 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 4 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 12.7.6",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|amd64|4|8|none|0|||"
          },
          "machine_class": "darwin|amd64|4|8|none|0|||",
          "machines": 1,
          "tokens_per_sec": 1.85,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 17.4407,
          "cost_breakdown": {
            "tokens_per_sec": 1.85,
            "usd_per_hour": 0.1165,
            "usd_per_million": 17.4407,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 16.1732
              },
              {
                "resource": "Memory",
                "quantity": "2.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.934,
                "usd_per_million": 1.0426
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.2248
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Core i5-5350U",
              "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
              "clock_ghz": 1.8,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 3,
              "single_core_hashes_per_sec": 93.7,
              "multi_core_hashes_per_sec": 195.1,
              "sustained_ratio": 0.7,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T23:57:35.534Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 8,
              "bandwidth_gb_s": 0.1,
              "latency_ns": 170.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T23:57:37.012Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 89.9,
              "write_mb_s": 44.3,
              "random_read_iops": 25045,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T23:57:40.536Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-16T00:27:19.106Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "tqZ75GjCOqnMPnoK21BpdHREBPxjFVQk64Hiwr5hdjyRsDYon+Wo+0UoXxKRWAsPVibx9SDEybqkagrb8BQyBw==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MS44NTQ5MTY2Nzc5MzM5MTY4LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5NDc4ODgzOTEwNjk4MDAwMCwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDg0MTQ2NGE0YTZkNDVhOGJkNzI5ZTAzMmI0YzFiMDNkOWFkNjlhNThjNzZlMGE1YzUxYmQwMjRiZmRkNjg2NzIiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImRhcndpbi9hbWQ2NC9hcHBsZS84ZyIsIm1vZGVsIjoiaGYuY28vdW5zbG90aC9QaGktNC1taW5pLXJlYXNvbmluZy1HR1VGOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50IjoxLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjY2My42NDE2NjksInRvdGFsX3Rva2VucyI6MTIzMX19",
            "payload_sha256": "kvdmc0V7b5auudEbUfuRL/S2TQKce7MmvP3YaEvzzGo="
          },
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W35"
          ],
          "providers": [
            "d68672"
          ]
        },
        {
          "machine": "Mac · no GPU · 4 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 4 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 12.7.6",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|amd64|4|8|none|0|||"
          },
          "machine_class": "darwin|amd64|4|8|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q6_K (6-bit) by lmstudio-community",
          "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q6_K",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Core i5-5350U",
              "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
              "clock_ghz": 1.8,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 3,
              "single_core_hashes_per_sec": 93.7,
              "multi_core_hashes_per_sec": 195.1,
              "sustained_ratio": 0.7,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T23:57:35.534Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 8,
              "bandwidth_gb_s": 0.1,
              "latency_ns": 170.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T23:57:37.012Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 89.9,
              "write_mb_s": 44.3,
              "random_read_iops": 25045,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T23:57:40.536Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-08-28T01:42:54.332Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/mini-reasoning/lmstudio-community-q6_k-ollama",
          "hub_url": "https://huggingface.co/lmstudio-community/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W35"
          ],
          "providers": [
            "d68672"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 8 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 8 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 15.7.3",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|8|8|apple|0|||"
          },
          "machine_class": "darwin|arm64|8|8|apple|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ1_S",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-08T10:28:59.157Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq1_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W37",
            "2026-W38",
            "2026-W41",
            "2026-W42"
          ],
          "providers": [
            "0c5265"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ1_S",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-11T18:33:38.903Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq1_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W37",
            "2026-W38",
            "2026-W41",
            "2026-W42"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 4 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 4 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 12.7.6",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|amd64|4|8|apple|0|||"
          },
          "machine_class": "darwin|amd64|4|8|apple|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ1_S",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Core i5-5350U",
              "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
              "clock_ghz": 1.8,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 3,
              "single_core_hashes_per_sec": 93.7,
              "multi_core_hashes_per_sec": 195.1,
              "sustained_ratio": 0.7,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T23:57:35.534Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 8,
              "bandwidth_gb_s": 0.1,
              "latency_ns": 170.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T23:57:37.012Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 89.9,
              "write_mb_s": 44.3,
              "random_read_iops": 25045,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T23:57:40.536Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-13T03:45:25.064Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq1_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W37",
            "2026-W38",
            "2026-W41",
            "2026-W42"
          ],
          "providers": [
            "d68672"
          ]
        },
        {
          "machine": "Windows · AMD 1 GB · 8 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · AMD 1 GB · 8 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.22631",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 7,
            "gpu_vendor": "AMD",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 1,
            "host": null,
            "vendor_org": "amd",
            "host_org": null,
            "key": "windows|amd64|8|7|amd|1|||"
          },
          "machine_class": "windows|amd64|8|7|amd|1|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ1_S",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 997.6,
              "multi_core_hashes_per_sec": 8516.9,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T08:14:32.559Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "memory": {
              "total_gb": 7,
              "bandwidth_gb_s": 6,
              "latency_ns": 61.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T08:14:33.251Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2347.3,
              "write_mb_s": 204.2,
              "random_read_iops": 96070,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-10-29T08:14:33.869Z",
              "measured_in": "IN_560001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-15T07:42:59.096Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq1_s-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W37",
            "2026-W38",
            "2026-W41",
            "2026-W42"
          ],
          "providers": [
            "d061c9"
          ]
        },
        {
          "machine": "Windows · AMD 1 GB · 8 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · AMD 1 GB · 8 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.22631",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 7,
            "gpu_vendor": "AMD",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 1,
            "host": null,
            "vendor_org": "amd",
            "host_org": null,
            "key": "windows|amd64|8|7|amd|1|||"
          },
          "machine_class": "windows|amd64|8|7|amd|1|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 4096,
          "build": "IQ3_XXS (3-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 997.6,
              "multi_core_hashes_per_sec": 8516.9,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T08:14:32.559Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "memory": {
              "total_gb": 7,
              "bandwidth_gb_s": 6,
              "latency_ns": 61.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T08:14:33.251Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2347.3,
              "write_mb_s": 204.2,
              "random_read_iops": 96070,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-10-29T08:14:33.869Z",
              "measured_in": "IN_560001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-15T08:17:39.721Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W41",
            "2026-W42"
          ],
          "providers": [
            "d061c9"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 8 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 8 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 15.7.3",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|8|8|apple|0|||"
          },
          "machine_class": "darwin|arm64|8|8|apple|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 4096,
          "build": "IQ3_XXS (3-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-08T11:03:39.716Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W41",
            "2026-W42"
          ],
          "providers": [
            "0c5265"
          ]
        },
        {
          "machine": "Mac · no GPU · 4 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 4 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 12.7.6",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|amd64|4|8|none|0|||"
          },
          "machine_class": "darwin|amd64|4|8|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q2_K (2-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q2_K",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Core i5-5350U",
              "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
              "clock_ghz": 1.8,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 3,
              "single_core_hashes_per_sec": 93.7,
              "multi_core_hashes_per_sec": 195.1,
              "sustained_ratio": 0.7,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T23:57:35.534Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 8,
              "bandwidth_gb_s": 0.1,
              "latency_ns": 170.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T23:57:37.012Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 89.9,
              "write_mb_s": 44.3,
              "random_read_iops": 25045,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T23:57:40.536Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-08-27T21:18:14.816Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/mini-reasoning/unsloth-q2_k-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-reasoning-gguf",
          "weeks": [
            "2026-W35"
          ],
          "providers": [
            "d68672"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/microsoft/Phi-4-mini-reasoning"
      },
      "builds": 10,
      "variants": [],
      "bottom_line": "189 tokens a second on Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Llama 3.2 1B",
      "made_by": "meta-llama",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/646cf8084eefb026fb8fd8bc/oCTqufkdTkjyGodsx1vo1.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/llama-3-2/1b",
      "hub_url": "https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct",
      "download_gb": 0.8,
      "params_b": 1.2,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 32768,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "English",
        "Hindi",
        "Dutch",
        "Tamil",
        "Telugu",
        "Kannada",
        "Malayalam",
        "Bengali"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Mac · no GPU · 10 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Mac · no GPU · 10 CPUs · 16 GB RAM",
          "os": "Mac",
          "os_version": "macOS 26.5.2",
          "cpus": 10,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": "apple",
          "host_org": null,
          "key": "darwin|arm64|10|16|none|0|||"
        },
        "machine_class": "darwin|arm64|10|16|none|0|||",
        "machines": 1,
        "tokens_per_sec": 113.27,
        "context_tokens": 8192,
        "build": "Q4_K_M (4-bit) by maziyarpanahi",
        "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.8056,
        "cost_breakdown": {
          "tokens_per_sec": 113.27,
          "usd_per_hour": 0.3285,
          "usd_per_million": 0.8056,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "9 of 10 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 90,
              "usd_per_million": 0.7946
            },
            {
              "resource": "Memory",
              "quantity": "0.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0,
              "usd_per_million": 0
            },
            {
              "resource": "Energy",
              "quantity": "45 W modelled (9 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 1.25,
              "usd_per_million": 0.011
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Apple M4",
            "model_raw": "Apple M4",
            "clock_ghz": null,
            "cores_logical": 10,
            "cores_physical": 10,
            "threads_per_core": 1,
            "sockets": 1,
            "l2_kb": 4096,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2850.2,
            "multi_core_hashes_per_sec": 22838.7,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-02T08:29:50.654Z",
            "measured_in": "IN_110054",
            "signed": true
          },
          "memory": {
            "total_gb": 16,
            "bandwidth_gb_s": 42.7,
            "latency_ns": 6.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-02T08:29:51.180Z",
            "measured_in": "IN_110054",
            "signed": true
          },
          "storage": {
            "read_mb_s": 13543.1,
            "write_mb_s": 1696,
            "random_read_iops": 1733096,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-02T08:29:51.427Z",
            "measured_in": "IN_110054",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-02T11:31:50.442Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
          "signature": "v8YfvttGpZh94elK5bm4mPYsZhsxopjsTz4SgFL6U06m8TRt5JHd/fVvwNAKdgjanGZnagDv1tHEPQ1DSTb/Ag==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTEzLjI2NjI0NjE1NTI0MDIsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzNjE5MTEwNDQyMTA5MDAwLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwYWQyYmExZmFiYmM4ZTVmMjM4OTJjMDEzNDIyNTc5MGRkMDM1ZDY1MmNlNTQxZGMxNmUwYjA5OGM1NmFmYmI4YSIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoiZGFyd2luL2FybTY0L2FwcGxlLzE2ZyIsIm1vZGVsIjoiaGYuY28vTWF6aXlhclBhbmFoaS9MbGFtYS0zLjItMUItSW5zdHJ1Y3QtR0dVRjpRNF9LX00iLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6MTAsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6Mi4zODM3NjQsInRvdGFsX3Rva2VucyI6MjcwfX0=",
          "payload_sha256": "uaDhtI2e7/M5CoNNcthDrIfB1IkMJcolt3e8NxLixKM="
        },
        "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/maziyarpanahi/llama-3.2-1b-instruct-gguf",
        "weeks": [
          "2026-W34",
          "2026-W35"
        ],
        "providers": [
          "afbb8a"
        ]
      },
      "cheapest": {
        "machine": "Mac · no GPU · 10 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Mac · no GPU · 10 CPUs · 16 GB RAM",
          "os": "Mac",
          "os_version": "macOS 26.5.2",
          "cpus": 10,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": "apple",
          "host_org": null,
          "key": "darwin|arm64|10|16|none|0|||"
        },
        "machine_class": "darwin|arm64|10|16|none|0|||",
        "machines": 1,
        "tokens_per_sec": 113.27,
        "context_tokens": 8192,
        "build": "Q4_K_M (4-bit) by maziyarpanahi",
        "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.8056,
        "cost_breakdown": {
          "tokens_per_sec": 113.27,
          "usd_per_hour": 0.3285,
          "usd_per_million": 0.8056,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "9 of 10 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 90,
              "usd_per_million": 0.7946
            },
            {
              "resource": "Memory",
              "quantity": "0.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0,
              "usd_per_million": 0
            },
            {
              "resource": "Energy",
              "quantity": "45 W modelled (9 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 1.25,
              "usd_per_million": 0.011
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Apple M4",
            "model_raw": "Apple M4",
            "clock_ghz": null,
            "cores_logical": 10,
            "cores_physical": 10,
            "threads_per_core": 1,
            "sockets": 1,
            "l2_kb": 4096,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2850.2,
            "multi_core_hashes_per_sec": 22838.7,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-02T08:29:50.654Z",
            "measured_in": "IN_110054",
            "signed": true
          },
          "memory": {
            "total_gb": 16,
            "bandwidth_gb_s": 42.7,
            "latency_ns": 6.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-02T08:29:51.180Z",
            "measured_in": "IN_110054",
            "signed": true
          },
          "storage": {
            "read_mb_s": 13543.1,
            "write_mb_s": 1696,
            "random_read_iops": 1733096,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-02T08:29:51.427Z",
            "measured_in": "IN_110054",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-02T11:31:50.442Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
          "signature": "v8YfvttGpZh94elK5bm4mPYsZhsxopjsTz4SgFL6U06m8TRt5JHd/fVvwNAKdgjanGZnagDv1tHEPQ1DSTb/Ag==",
          "signing_payload_b64": "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",
          "payload_sha256": "uaDhtI2e7/M5CoNNcthDrIfB1IkMJcolt3e8NxLixKM="
        },
        "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/maziyarpanahi/llama-3.2-1b-instruct-gguf",
        "weeks": [
          "2026-W34",
          "2026-W35"
        ],
        "providers": [
          "afbb8a"
        ]
      },
      "machine_kinds": 8,
      "machines_total": 22,
      "weeks_measured": [
        "2026-W34",
        "2026-W35",
        "2026-W36",
        "2026-W37",
        "2026-W38",
        "2026-W39",
        "2026-W40",
        "2026-W41",
        "2026-W42",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Mac · no GPU · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|none|0|||"
          },
          "machine_class": "darwin|arm64|10|16|none|0|||",
          "machines": 1,
          "tokens_per_sec": 113.27,
          "context_tokens": 8192,
          "build": "Q4_K_M (4-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.8056,
          "cost_breakdown": {
            "tokens_per_sec": 113.27,
            "usd_per_hour": 0.3285,
            "usd_per_million": 0.8056,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 0.7946
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "45 W modelled (9 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 1.25,
                "usd_per_million": 0.011
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "v8YfvttGpZh94elK5bm4mPYsZhsxopjsTz4SgFL6U06m8TRt5JHd/fVvwNAKdgjanGZnagDv1tHEPQ1DSTb/Ag==",
            "signing_payload_b64": "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",
            "payload_sha256": "uaDhtI2e7/M5CoNNcthDrIfB1IkMJcolt3e8NxLixKM="
          },
          "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/llama-3.2-1b-instruct-gguf",
          "weeks": [
            "2026-W34",
            "2026-W35"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 8 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 8 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 15.7.3",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|8|8|apple|0|||"
          },
          "machine_class": "darwin|arm64|8|8|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 20.37,
          "context_tokens": 32768,
          "build": null,
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "cost_usd_per_million": 8.5986,
          "cost_breakdown": {
            "tokens_per_sec": 20.37,
            "usd_per_hour": 0.6305,
            "usd_per_million": 8.5986,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 3.4367
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 4.9096
              },
              {
                "resource": "Energy",
                "quantity": "185 W modelled (7 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.139,
                "usd_per_million": 0.2523
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-08T15:04:04.375Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ab92819c6391e87d553e27cec9a18a52937a84a123bf752d00c2aa2bee0548a8",
            "signature": "eebpktI5uYJizQhh7Sn0Iv+0jvsz9CWBbeHIi9IeoqUpY6GQXP6bH4ZrCTw4e9xfcfjk3Uz2q9+r0kPMHICLAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "qVEsr7bcara1XYeXsf0n2J0VT2RIdNte7F5A6GDGXWE="
          },
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "hub_url": null,
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "0548a8"
          ]
        },
        {
          "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "windows|amd64|8|16|none|0|||"
          },
          "machine_class": "windows|amd64|8|16|none|0|||",
          "machines": 1,
          "tokens_per_sec": 17.72,
          "context_tokens": 4096,
          "build": "IQ3_XS (3-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:IQ3_XS",
          "engine": "ollama",
          "cost_usd_per_million": 4.0054,
          "cost_breakdown": {
            "tokens_per_sec": 17.72,
            "usd_per_hour": 0.2555,
            "usd_per_million": 4.0054,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 3.9505
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0549
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "WqKWZGWm3xXMnhC4j4wDAQTBg1V3Xz930DW5aKw9lPqBOK0RGyn8kqgRg/OAMw+zHviu9JeGTlmk3/QlaIgvDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "HHahuvMAN84Rid0xaed1SEw5x/umOUQUKwUaH/9YE3E="
          },
          "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-iq3_xs-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/llama-3.2-1b-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W37"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 12.24,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.5368,
          "cost_breakdown": {
            "tokens_per_sec": 12.24,
            "usd_per_hour": 0.1118,
            "usd_per_million": 2.5368,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.4515
              },
              {
                "resource": "Memory",
                "quantity": "0.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.627,
                "usd_per_million": 0.0512
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0341
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "pd7F5+EgRJwYCbecjdb4Rie4hVhcSZJQR76WRWXwVJ832fw1PUpZTJ5WZ/aGo3bYLeKVbQ/1W30O+1LmBYqnBg==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTIuMjM3NTg4ODg4MTIzMjA4LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5MzQwNTkxNzI5ODY0ODg5MSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMGVlMzVkYzQ2YTIyMTJjYjMxNDQ0Y2E0YzE5NjY5YWM4ZDk3YjA5MjUyMGI4YzllNGQyOWFiNDhkOWVlNjFkM2IiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L25vbmUvMzJnIiwibW9kZWwiOiJoZi5jby9iYXJ0b3dza2kvTGxhbWEtMy4yLTFCLUluc3RydWN0LUdHVUY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjUyLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjIwOC43MDEyNDE5OTcsInRvdGFsX3Rva2VucyI6MjU1NH19",
            "payload_sha256": "fzkWkaVB3SPSyX/ndXuuGiweZ7ZMM7nskja753MYi9M="
          },
          "build_page": "/inferences/llama-3-2/1b/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-1b-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W37",
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "windows|amd64|8|16|none|0|||"
          },
          "machine_class": "windows|amd64|8|16|none|0|||",
          "machines": 1,
          "tokens_per_sec": 11.62,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 6.161,
          "cost_breakdown": {
            "tokens_per_sec": 11.62,
            "usd_per_hour": 0.2578,
            "usd_per_million": 6.161,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 6.0234
              },
              {
                "resource": "Memory",
                "quantity": "0.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.627,
                "usd_per_million": 0.054
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0836
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "wDTHJzoIikB2dE6nDNndfzo2VtP9Jhv7K3dpiJmPAnaypEJrW8jge5+iLBanTuViujbRJxKWT1o3DkQDLtHsCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "uUiXSylyDMFzJvJCGvFj4j9UR78WmdJTQJwGMDqTtC0="
          },
          "build_page": "/inferences/llama-3-2/1b/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-1b-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W37",
            "2026-W44"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 3,
          "tokens_per_sec": 11.45,
          "context_tokens": 32768,
          "build": null,
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.6566,
          "cost_breakdown": {
            "tokens_per_sec": 11.45,
            "usd_per_hour": 0.1095,
            "usd_per_million": 2.6566,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.6202
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0364
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T06:35:24.096Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "VwsbNGAF8UjFxH5JNbn4FCMEzEO4kR9zn//+9cHV5sEi0nm5UvPBDwURrb4XElNl1cc5+12lUWkUOJsX0y2aCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "qrUwcerxoHklzLicmUtNKQ8zcVGaGQaMF19KWfgt82o="
          },
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "hub_url": null,
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e3654f",
            "e61d3b",
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 11.25,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by hugging-quants",
          "ref": "hf.co/hugging-quants/Llama-3.2-1B-Instruct-Q4_K_M-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.7606,
          "cost_breakdown": {
            "tokens_per_sec": 11.25,
            "usd_per_hour": 0.1118,
            "usd_per_million": 2.7606,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.6677
              },
              {
                "resource": "Memory",
                "quantity": "0.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.627,
                "usd_per_million": 0.0558
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0371
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "fwWkqvBRrU4a78QT3IF6wnRaEuR1/94JWriCWzNq90enQ+WTd4zXOWK4Y5fgAlQ/Ua0iX1fYtUsEI/ieMlKNAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "gmkdtergC55JVGdKpWdG0gNXdvS8Yl6YsEV/qvCCYaU="
          },
          "build_page": "/inferences/llama-3-2/1b~hugging-quants-llama-3-2-k-m/hugging-quants-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/hugging-quants/llama-3.2-1b-instruct-q4_k_m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 3,
          "tokens_per_sec": 10.79,
          "context_tokens": 8192,
          "build": null,
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.8196,
          "cost_breakdown": {
            "tokens_per_sec": 10.79,
            "usd_per_hour": 0.1095,
            "usd_per_million": 2.8196,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.781
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0387
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 185.9,
              "multi_core_hashes_per_sec": 588,
              "sustained_ratio": 1.4,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-14T19:37:12.663Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 1.7,
              "latency_ns": 71.3,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-14T19:37:15.832Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2202.4,
              "write_mb_s": 15.5,
              "random_read_iops": 196685,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-14T19:37:24.078Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T20:22:46.948Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220dcdccbd7ca3461f853cbf5c0f75217e5f90c96896457d92b59bed1b6ee4e94b2",
            "signature": "YnmWexpPC8JA+V6mE8GoELi444EXbvlieEu8EMkIs3sIi9Ru/YJxvPVDAp2KeeN7xyGH0bGJWR8ATiOhrJ5rBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "9/LgoM2fHYoSl4S+4c/CUnKMNv0+w53a/nJDIz2a6cE="
          },
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "hub_url": null,
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "4e94b2",
            "662a4e",
            "73d610"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 10.3,
          "context_tokens": null,
          "build": "Q3_K_L (3-bit) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "cost_usd_per_million": 3.007,
          "cost_breakdown": {
            "tokens_per_sec": 10.3,
            "usd_per_hour": 0.1115,
            "usd_per_million": 3.007,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.9114
              },
              {
                "resource": "Memory",
                "quantity": "0.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.568,
                "usd_per_million": 0.0551
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0405
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "y3BjTARSXdwp5GFW5MZgPMg80ndl98HpL8EynpK2VZh5x6CWI0x6Vu/2kydrSuJ0mpE/PvJPg3ajE2jMy0LvDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "gujLB37UrpZPYg5xdCMPUao7DFHZJGZEgAZyHvZdHpw="
          },
          "build_page": "/inferences/llama-3-2/1b/bartowski-q3_k_l-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-1b-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 8.46,
          "context_tokens": 4096,
          "build": "Q8_0 (8-bit) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 3.717,
          "cost_breakdown": {
            "tokens_per_sec": 8.46,
            "usd_per_hour": 0.1132,
            "usd_per_million": 3.717,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 3.5465
              },
              {
                "resource": "Memory",
                "quantity": "1.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.025,
                "usd_per_million": 0.1212
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0493
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "EOD1LKNsSNZ16rOZ1H2o0TdhPtv6al2Pd8YwBzwROpz+9QI7eLVZst9csdiNzJCwk8qe3UgCYPbaxXHAQnBuCg==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6OC40NTg5ODgyNDI1NDAyMjIsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzNDA1OTE3Mjk4NjQ4ODkxLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwZWUzNWRjNDZhMjIxMmNiMzE0NDRjYTRjMTk2NjlhYzhkOTdiMDkyNTIwYjhjOWU0ZDI5YWI0OGQ5ZWU2MWQzYiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbm9uZS8zMmciLCJtb2RlbCI6ImhmLmNvL2JhcnRvd3NraS9MbGFtYS0zLjItMUItSW5zdHJ1Y3QtR0dVRjpROF8wIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjU4LCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjM3Ny4zNTAwOTI5OTg5OTk5NywidG90YWxfdG9rZW5zIjozMTkyfX0=",
            "payload_sha256": "M0zi1Gt9fv2vG0Lq37aDGZxMdA5KHkN4yTta1kv2Rs0="
          },
          "build_page": "/inferences/llama-3-2/1b/bartowski-q8_0-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-1b-instruct-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 6.6,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:IQ1_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 4.6098,
          "cost_breakdown": {
            "tokens_per_sec": 6.6,
            "usd_per_hour": 0.1095,
            "usd_per_million": 4.6098,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 4.5466
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0632
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "YkH8hjmB0D1RxTb0myTIO3rpNHcKDliF1WDro1t+hjnHFw97QCDFGw/BUbLpk7NZV43F1I7dAfklAXD0xaoZBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "l/x+eT5oT+zLWaIe2WjY8QxKTNUTZqNv7zx60d1ALWA="
          },
          "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-iq1_s-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/llama-3.2-1b-instruct-gguf",
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W37",
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.22631",
            "cpus": 8,
            "cpu_model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
            "ram_gb": 7,
            "gpu_vendor": "AMD",
            "gpu_model": "AMD Radeon(TM) Vega 8 Mobile Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "amd",
            "host_org": null,
            "key": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 5.62,
          "context_tokens": 4096,
          "build": null,
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "cost_usd_per_million": 31.1559,
          "cost_breakdown": {
            "tokens_per_sec": 5.62,
            "usd_per_hour": 0.6305,
            "usd_per_million": 31.1559,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 12.4525
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 17.7893
              },
              {
                "resource": "Energy",
                "quantity": "185 W modelled (7 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.139,
                "usd_per_million": 0.9142
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 997.6,
              "multi_core_hashes_per_sec": 8516.9,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T08:14:32.559Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "memory": {
              "total_gb": 7,
              "bandwidth_gb_s": 6,
              "latency_ns": 61.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T08:14:33.251Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2347.3,
              "write_mb_s": 204.2,
              "random_read_iops": 96070,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-10-29T08:14:33.869Z",
              "measured_in": "IN_560001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-29T09:04:03.140Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
            "signature": "2Q//JA9gQANmknVAW7mbRxfmRtQbcA2w0mdGysoT3hwNn/ViPWMv1vhtQFw+RjkgZbYUF7vIjOn1fWsujlBqDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "F1N2EFfzZqANMjAs9vMbJvh7roRWtG0s4rOX+imwhCI="
          },
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "hub_url": null,
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "d061c9"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 5.28,
          "context_tokens": null,
          "build": "IQ3_M (3-bit) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:IQ3_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 5.862,
          "cost_breakdown": {
            "tokens_per_sec": 5.28,
            "usd_per_hour": 0.1113,
            "usd_per_million": 5.862,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 5.6865
              },
              {
                "resource": "Memory",
                "quantity": "0.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.509,
                "usd_per_million": 0.0965
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.079
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "bIEnv7GZNBLKxTlX0PehieTL3iLHvTLkO+GhNpz73QZ2inxjwS25vaLTi7Y0gutWwcBMsfOy4z+/KqRxHiKYAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "L4owj79V9xN0X3NttAyRjeSHHK01XGMNfVAWdbbO1ms="
          },
          "build_page": "/inferences/llama-3-2/1b/bartowski-iq3_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-1b-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Mac · no GPU · 4 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 4 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 12.7.6",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|amd64|4|8|none|0|||"
          },
          "machine_class": "darwin|amd64|4|8|none|0|||",
          "machines": 1,
          "tokens_per_sec": 4.73,
          "context_tokens": 16384,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 6.5625,
          "cost_breakdown": {
            "tokens_per_sec": 4.73,
            "usd_per_hour": 0.1118,
            "usd_per_million": 6.5625,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.3418
              },
              {
                "resource": "Memory",
                "quantity": "0.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.627,
                "usd_per_million": 0.1325
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0882
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Core i5-5350U",
              "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
              "clock_ghz": 1.8,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 3,
              "single_core_hashes_per_sec": 93.7,
              "multi_core_hashes_per_sec": 195.1,
              "sustained_ratio": 0.7,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T23:57:35.534Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 8,
              "bandwidth_gb_s": 0.1,
              "latency_ns": 170.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T23:57:37.012Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 89.9,
              "write_mb_s": 44.3,
              "random_read_iops": 25045,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T23:57:40.536Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-16T00:27:19.106Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "xeXVtalj4tfLU93gEx2RKmQEB7H0C+MCNagv4wmgc6gaPAft5MrDgo3e1NC44pYDrj/AGC72HJZ574zmotUvBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "7cxiDyJf/0sZM6JbVuSqShplnyJrTQQjL6E2BaTbPfk="
          },
          "build_page": "/inferences/llama-3-2/1b/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-1b-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W37",
            "2026-W44"
          ],
          "providers": [
            "d68672"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 4.3,
          "context_tokens": null,
          "build": "F16 (16-bit, full precision) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:F16",
          "engine": "ollama-host",
          "cost_usd_per_million": 7.5285,
          "cost_breakdown": {
            "tokens_per_sec": 4.3,
            "usd_per_hour": 0.1164,
            "usd_per_million": 7.5285,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.9836
              },
              {
                "resource": "Memory",
                "quantity": "2.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.924,
                "usd_per_million": 0.4479
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0971
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "OcsybWnv9AQwqxC0kOi5jO9i+oxUXhvxsnlYtvieoO1uZ44LaWmSAWeSqkRqkySdAW7dkiv4iuSPN+ixO72MAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "/CbkYXkRopu7qgHrvn1xgFSmnOxl4/NX/aDquw//4AE="
          },
          "build_page": "/inferences/llama-3-2/1b/bartowski-f16-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-1b-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 4 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 4 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 12.7.6",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|amd64|4|8|apple|0|||"
          },
          "machine_class": "darwin|amd64|4|8|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 3.37,
          "context_tokens": 32768,
          "build": null,
          "ref": "llama3.2:1b",
          "engine": "ollama-host",
          "cost_usd_per_million": 39.9708,
          "cost_breakdown": {
            "tokens_per_sec": 3.37,
            "usd_per_hour": 0.4845,
            "usd_per_million": 39.9708,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 8.9099
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 29.6998
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 1.3611
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Core i5-5350U",
              "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
              "clock_ghz": 1.8,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 3,
              "single_core_hashes_per_sec": 93.7,
              "multi_core_hashes_per_sec": 195.1,
              "sustained_ratio": 0.7,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T23:57:35.534Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 8,
              "bandwidth_gb_s": 0.1,
              "latency_ns": 170.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T23:57:37.012Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 89.9,
              "write_mb_s": 44.3,
              "random_read_iops": 25045,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T23:57:40.536Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-16T00:27:19.106Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "xd8hLSzOrPU/DLkbdsBtG/33kOLVd+irZEX/nDwAgKZcgPdOm4RD8XlZyLXQSZnH6v7rUNnqstW4HoCUu4e8DQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "aRps7eQcPSmCtn4JLzuz5SWTh+NJhBImq4KD5mhG5nw="
          },
          "build_page": "/inferences/llama-3-2/1b/ollama-q8_0-ollama",
          "hub_url": null,
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "d68672"
          ]
        },
        {
          "machine": "Mac · no GPU · 4 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 4 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 12.7.6",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|amd64|4|8|none|0|||"
          },
          "machine_class": "darwin|amd64|4|8|none|0|||",
          "machines": 1,
          "tokens_per_sec": 3.26,
          "context_tokens": null,
          "build": "IQ1_S (1-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:IQ1_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 9.3281,
          "cost_breakdown": {
            "tokens_per_sec": 3.26,
            "usd_per_hour": 0.1095,
            "usd_per_million": 9.3281,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 9.2002
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1279
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Core i5-5350U",
              "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
              "clock_ghz": 1.8,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 3,
              "single_core_hashes_per_sec": 93.7,
              "multi_core_hashes_per_sec": 195.1,
              "sustained_ratio": 0.7,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T23:57:35.534Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 8,
              "bandwidth_gb_s": 0.1,
              "latency_ns": 170.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T23:57:37.012Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 89.9,
              "write_mb_s": 44.3,
              "random_read_iops": 25045,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T23:57:40.536Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-16T00:27:19.106Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "SpW5/M0bkEYuxG2s6CD/eUez5QpwFbFF8abJOI0SF2/6IVIyygTNqXo3Tf6PdCnAJdN7xNS+4i6M/bbliZekCg==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6My4yNjA4MDgxNTM5OTg2NjQ0LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5NDc4ODgzOTEwNjk4MDAwMCwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDg0MTQ2NGE0YTZkNDVhOGJkNzI5ZTAzMmI0YzFiMDNkOWFkNjlhNThjNzZlMGE1YzUxYmQwMjRiZmRkNjg2NzIiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImRhcndpbi9hbWQ2NC9hcHBsZS84ZyIsIm1vZGVsIjoiaGYuY28vTWF6aXlhclBhbmFoaS9MbGFtYS0zLjItMUItSW5zdHJ1Y3QtR0dVRjpJUTFfUyIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50IjoxLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjExOS4yOTU1OCwidG90YWxfdG9rZW5zIjozODl9fQ==",
            "payload_sha256": "qP0yV3jyFo3b5KApQMJNiR7uuvjnVSOlK19XpzQJGHA="
          },
          "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-iq1_s-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/llama-3.2-1b-instruct-gguf",
          "weeks": [
            "2026-W34",
            "2026-W35",
            "2026-W37",
            "2026-W44"
          ],
          "providers": [
            "d68672"
          ]
        },
        {
          "machine": "Mac · no GPU · 4 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 4 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 12.7.6",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|amd64|4|8|none|0|||"
          },
          "machine_class": "darwin|amd64|4|8|none|0|||",
          "machines": 1,
          "tokens_per_sec": 1.48,
          "context_tokens": 4096,
          "build": "IQ3_XS (3-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:IQ3_XS",
          "engine": "ollama-host",
          "cost_usd_per_million": 20.6089,
          "cost_breakdown": {
            "tokens_per_sec": 1.48,
            "usd_per_hour": 0.1095,
            "usd_per_million": 20.6089,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 20.3263
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.2825
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Core i5-5350U",
              "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
              "clock_ghz": 1.8,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 3,
              "single_core_hashes_per_sec": 93.7,
              "multi_core_hashes_per_sec": 195.1,
              "sustained_ratio": 0.7,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T23:57:35.534Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 8,
              "bandwidth_gb_s": 0.1,
              "latency_ns": 170.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T23:57:37.012Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 89.9,
              "write_mb_s": 44.3,
              "random_read_iops": 25045,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T23:57:40.536Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-16T00:27:19.106Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "JfIHWmj1EH4cJR9v4NAu/XbQRQSJKHXHnPKxT31jtg00FHAT+3ryZu/4Qonsex0udIs5ksxpg9gcpwwbVrZPDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "v7HypU3osEpbQ7DnJilzW9FaIvs16oSZ859bh3Ybhyg="
          },
          "build_page": "/inferences/llama-3-2/1b/maziyarpanahi-iq3_xs-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/llama-3.2-1b-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W37"
          ],
          "providers": [
            "d68672"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "meta-llama/llama-3.2-1b-instruct",
        "url": "https://openrouter.ai/meta-llama/llama-3.2-1b-instruct",
        "input_usd_per_million": 0.027,
        "output_usd_per_million": 0.201,
        "free_tier": false,
        "context_tokens": 60000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.027,
            "usd_per_m_out": 0.201,
            "context_length": 60000,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "Llama 3.2 community licence",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/meta-llama/Llama-3.2-1B-Instruct"
      },
      "builds": 11,
      "variants": [],
      "bottom_line": "113 tokens a second on Mac · no GPU · 10 CPUs · 16 GB RAM. About $0.81 per million tokens here, against OpenRouter's $0.20: the API is 4.0× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "MiniCPM-V 4 Base",
      "made_by": "openbmb",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1670387859384-633fe7784b362488336bbfad.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/minicpm-v-4/base",
      "hub_url": "https://huggingface.co/openbmb/MiniCPM-V-4",
      "download_gb": 2,
      "params_b": 3.6,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "returns valid JSON",
        "long answers",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 96.71,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM-V-4-gguf:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.2478,
        "cost_breakdown": {
          "tokens_per_sec": 96.71,
          "usd_per_hour": 0.7826,
          "usd_per_million": 2.2478,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 1.1374
            },
            {
              "resource": "Memory",
              "quantity": "2.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.699,
              "usd_per_million": 0.0176
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 1.034
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0589
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "1zm49qNy9FzO+ULoa6Xue2ZXqidyUIT8SRPzUXTgxVSTuxOglEkfOSenUcSJQ0gz0CEi9DBVrQLLSaNJGe8/BA==",
          "signing_payload_b64": "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",
          "payload_sha256": "YpGidNliB60trWm3kXZb/FY3+KBbBNf+yhuPj+vM448="
        },
        "build_page": "/inferences/minicpm-v-4/base/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm-v-4-gguf",
        "weeks": [
          "2026-W44",
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 96.71,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM-V-4-gguf:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.2478,
        "cost_breakdown": {
          "tokens_per_sec": 96.71,
          "usd_per_hour": 0.7826,
          "usd_per_million": 2.2478,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 1.1374
            },
            {
              "resource": "Memory",
              "quantity": "2.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.699,
              "usd_per_million": 0.0176
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 1.034
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0589
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "1zm49qNy9FzO+ULoa6Xue2ZXqidyUIT8SRPzUXTgxVSTuxOglEkfOSenUcSJQ0gz0CEi9DBVrQLLSaNJGe8/BA==",
          "signing_payload_b64": "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",
          "payload_sha256": "YpGidNliB60trWm3kXZb/FY3+KBbBNf+yhuPj+vM448="
        },
        "build_page": "/inferences/minicpm-v-4/base/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm-v-4-gguf",
        "weeks": [
          "2026-W44",
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 3,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W44",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 96.71,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM-V-4-gguf:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.2478,
          "cost_breakdown": {
            "tokens_per_sec": 96.71,
            "usd_per_hour": 0.7826,
            "usd_per_million": 2.2478,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 1.1374
              },
              {
                "resource": "Memory",
                "quantity": "2.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.699,
                "usd_per_million": 0.0176
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.034
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0589
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "1zm49qNy9FzO+ULoa6Xue2ZXqidyUIT8SRPzUXTgxVSTuxOglEkfOSenUcSJQ0gz0CEi9DBVrQLLSaNJGe8/BA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6OTYuNzE0ODY1NzY4MTQwNSwidW5pdCI6InRva2Vucy9zIiwidHMiOjE3OTQwMjgxOTI5NzA5NzIyMDUsIm5vZGVfZGlkIjoiZGlkOmVwbjowMDI0MDgwMTEyMjA1MjQwNTBlOTI5NmQzMTJjZDI4MjhjNDI2MTY1YmZmMjNiYWVlZjc2OWFkYTk3MDgzMDk4ODMyZjA2ZGFlNjc5IiwiZXh0cmEiOnsiaGFyZHdhcmVfY2xhc3MiOiJsaW51eC9hbWQ2NC9udmlkaWEvNjRnIiwibW9kZWwiOiJoZi5jby9vcGVuYm1iL01pbmlDUE0tVi00LWdndWY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjUxLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjExNi41NDg3ODQsInRvdGFsX3Rva2VucyI6MTEyNzJ9fQ==",
            "payload_sha256": "YpGidNliB60trWm3kXZb/FY3+KBbBNf+yhuPj+vM448="
          },
          "build_page": "/inferences/minicpm-v-4/base/openbmb-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm-v-4-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.22631",
            "cpus": 8,
            "cpu_model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
            "ram_gb": 7,
            "gpu_vendor": "AMD",
            "gpu_model": "AMD Radeon(TM) Vega 8 Mobile Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "amd",
            "host_org": null,
            "key": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 3.73,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM-V-4-gguf:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 47.3964,
          "cost_breakdown": {
            "tokens_per_sec": 3.73,
            "usd_per_hour": 0.6366,
            "usd_per_million": 47.3964,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 18.7615
              },
              {
                "resource": "Memory",
                "quantity": "2.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.699,
                "usd_per_million": 0.4554
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 26.8021
              },
              {
                "resource": "Energy",
                "quantity": "185 W modelled (7 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.139,
                "usd_per_million": 1.3774
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 997.6,
              "multi_core_hashes_per_sec": 8516.9,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T08:14:32.559Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "memory": {
              "total_gb": 7,
              "bandwidth_gb_s": 6,
              "latency_ns": 61.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T08:14:33.251Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2347.3,
              "write_mb_s": 204.2,
              "random_read_iops": 96070,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-10-29T08:14:33.869Z",
              "measured_in": "IN_560001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-29T09:04:03.140Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
            "signature": "XNSd6EbZlPSEliFS/q7+ArudNTCRwnpJidH8e0SNBk+HWVAJosM9Rd0X3Jjn/TKqviPx/JhrgkEJAhJYo2kbDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "/idEGoz2O6T6vdumwKuUNxUKFZqdTANGKDjFLz3SFto="
          },
          "build_page": "/inferences/minicpm-v-4/base/openbmb-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm-v-4-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "d061c9"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 3.21,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM-V-4-gguf:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 22.6155,
          "cost_breakdown": {
            "tokens_per_sec": 3.21,
            "usd_per_hour": 0.2616,
            "usd_per_million": 22.6155,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 21.7843
              },
              {
                "resource": "Memory",
                "quantity": "2.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.699,
                "usd_per_million": 0.5287
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.3025
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "FBPpsFx6elyeHQSgjNz2YqfZeKDPdQmLtMwsDBzpz5uXHnSbkxmX6bHyffYi9cRpAEyJYrGl1/N+tkB9bOvpBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "S/lbERe6+UhTzqbM5Wndafz4ZUsdRC4MrTfglEf4q5U="
          },
          "build_page": "/inferences/minicpm-v-4/base/openbmb-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm-v-4-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "2e5c02"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openbmb/MiniCPM-V-4"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "97 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Granite 4.1 3B",
      "made_by": "ibm-granite",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/639bcaa2445b133a4e942436/CEW-OjXkRkDNmTxSu8Egh.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/granite-4-1/3b",
      "hub_url": "https://huggingface.co/ibm-granite/granite-4.1-3b",
      "download_gb": 2,
      "params_b": 3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 65536,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "multi-step tool use",
        "writes code",
        "returns valid JSON",
        "long answers"
      ],
      "languages_passed": [
        "English",
        "Kannada",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 18.15,
        "context_tokens": 65536,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/granite-4.1-3b-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 11.9748,
        "cost_breakdown": {
          "tokens_per_sec": 18.15,
          "usd_per_hour": 0.7824,
          "usd_per_million": 11.9748,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 6.0611
            },
            {
              "resource": "Memory",
              "quantity": "2.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.629,
              "usd_per_million": 0.0898
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 5.5101
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.3137
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "Kks22pwqcoELK1LJREfFD+zJ3kmp/syvbvOSE1iHEUlb3XL/3DgLGzgAouogu5D6khef2xFTWvrvCk5k63CEAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "Nmi0mfb/kuXpYt4fsQv9/mlpRMKoACofADmppw0PhlY="
        },
        "build_page": "/inferences/granite-4-1/3b/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/granite-4.1-3b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 18.15,
        "context_tokens": 65536,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/granite-4.1-3b-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 11.9748,
        "cost_breakdown": {
          "tokens_per_sec": 18.15,
          "usd_per_hour": 0.7824,
          "usd_per_million": 11.9748,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 6.0611
            },
            {
              "resource": "Memory",
              "quantity": "2.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.629,
              "usd_per_million": 0.0898
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 5.5101
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.3137
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "Kks22pwqcoELK1LJREfFD+zJ3kmp/syvbvOSE1iHEUlb3XL/3DgLGzgAouogu5D6khef2xFTWvrvCk5k63CEAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "Nmi0mfb/kuXpYt4fsQv9/mlpRMKoACofADmppw0PhlY="
        },
        "build_page": "/inferences/granite-4-1/3b/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/granite-4.1-3b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 18.15,
          "context_tokens": 65536,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/granite-4.1-3b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 11.9748,
          "cost_breakdown": {
            "tokens_per_sec": 18.15,
            "usd_per_hour": 0.7824,
            "usd_per_million": 11.9748,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 6.0611
              },
              {
                "resource": "Memory",
                "quantity": "2.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.629,
                "usd_per_million": 0.0898
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 5.5101
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.3137
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "Kks22pwqcoELK1LJREfFD+zJ3kmp/syvbvOSE1iHEUlb3XL/3DgLGzgAouogu5D6khef2xFTWvrvCk5k63CEAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "Nmi0mfb/kuXpYt4fsQv9/mlpRMKoACofADmppw0PhlY="
          },
          "build_page": "/inferences/granite-4-1/3b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/granite-4.1-3b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/ibm-granite/granite-4.1-3b"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "18 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Ministral 3 3B-Instruct-2512",
      "made_by": "mistralai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/634c17653d11eaedd88b314d/9OgyfKstSZtbmsmuG8MbU.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/ministral-3/3b-instruct-2512",
      "hub_url": "https://huggingface.co/mistralai/Ministral-3-3B-Instruct-2512",
      "download_gb": 2,
      "params_b": 3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 65536,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "long answers",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [
        "English",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 16.51,
        "context_tokens": 65536,
        "build": "Q4_K_M (4-bit) by mistralai",
        "ref": "hf.co/mistralai/Ministral-3-3B-Instruct-2512-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 13.1616,
        "cost_breakdown": {
          "tokens_per_sec": 16.51,
          "usd_per_hour": 0.7825,
          "usd_per_million": 13.1616,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 6.6607
            },
            {
              "resource": "Memory",
              "quantity": "2.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.666,
              "usd_per_million": 0.1009
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 6.0552
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.3448
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "M9RX6WM7k/iuAUJBwS3JEJChcUEJjDx8CQpa9aw+p39b9GuIjQXlqHLxpVf8AzICNErJY0ImEFmR7i4L8XOQBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "VivKeR6liU0dRGiOvj7R5OPjDCMZqoxAQEwngG/j88w="
        },
        "build_page": "/inferences/ministral-3/3b-instruct-2512/mistralai-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/mistralai/ministral-3-3b-instruct-2512-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 16.51,
        "context_tokens": 65536,
        "build": "Q4_K_M (4-bit) by mistralai",
        "ref": "hf.co/mistralai/Ministral-3-3B-Instruct-2512-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 13.1616,
        "cost_breakdown": {
          "tokens_per_sec": 16.51,
          "usd_per_hour": 0.7825,
          "usd_per_million": 13.1616,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 6.6607
            },
            {
              "resource": "Memory",
              "quantity": "2.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.666,
              "usd_per_million": 0.1009
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 6.0552
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.3448
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "M9RX6WM7k/iuAUJBwS3JEJChcUEJjDx8CQpa9aw+p39b9GuIjQXlqHLxpVf8AzICNErJY0ImEFmR7i4L8XOQBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "VivKeR6liU0dRGiOvj7R5OPjDCMZqoxAQEwngG/j88w="
        },
        "build_page": "/inferences/ministral-3/3b-instruct-2512/mistralai-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/mistralai/ministral-3-3b-instruct-2512-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 2,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 16.51,
          "context_tokens": 65536,
          "build": "Q4_K_M (4-bit) by mistralai",
          "ref": "hf.co/mistralai/Ministral-3-3B-Instruct-2512-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 13.1616,
          "cost_breakdown": {
            "tokens_per_sec": 16.51,
            "usd_per_hour": 0.7825,
            "usd_per_million": 13.1616,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 6.6607
              },
              {
                "resource": "Memory",
                "quantity": "2.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.666,
                "usd_per_million": 0.1009
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 6.0552
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.3448
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "M9RX6WM7k/iuAUJBwS3JEJChcUEJjDx8CQpa9aw+p39b9GuIjQXlqHLxpVf8AzICNErJY0ImEFmR7i4L8XOQBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "VivKeR6liU0dRGiOvj7R5OPjDCMZqoxAQEwngG/j88w="
          },
          "build_page": "/inferences/ministral-3/3b-instruct-2512/mistralai-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/mistralai/ministral-3-3b-instruct-2512-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 4.66,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by mistralai",
          "ref": "hf.co/mistralai/Ministral-3-3B-Instruct-2512-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 15.5973,
          "cost_breakdown": {
            "tokens_per_sec": 4.66,
            "usd_per_hour": 0.2615,
            "usd_per_million": 15.5973,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 15.0308
              },
              {
                "resource": "Memory",
                "quantity": "2.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.666,
                "usd_per_million": 0.3577
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.2087
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "ykNQ80DXwnULjVCK1xBzggzo9FiwauQOKZkox0ur+/MQcvLjFIYmSOC8ec0qlLLx6QDe0C76GH3QeSe/SwI7AQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "uEEtDl6YSJa7MhAkqO5u4jPkUP8DiRMAl9Fi3OT+BGM="
          },
          "build_page": "/inferences/ministral-3/3b-instruct-2512/mistralai-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/mistralai/ministral-3-3b-instruct-2512-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "2e5c02"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "mistralai/ministral-3b-2512",
        "url": "https://openrouter.ai/mistralai/ministral-3b-2512",
        "input_usd_per_million": 0.1,
        "output_usd_per_million": 0.1,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.1,
            "context_length": 131072,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 0.11,
            "usd_per_m_out": 0.11,
            "context_length": 131072,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/mistralai/Ministral-3-3B-Instruct-2512"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "17 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. About $13 per million tokens here, against OpenRouter's $0.10: the API is 132× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "Hunyuan MT 2 30B-A3B",
      "made_by": "tencent",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/Lp3m-XLpjQGwBItlvn69q.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/hy-mt-2/30b-a3b",
      "hub_url": "https://huggingface.co/tencent/Hy-MT2-30B-A3B",
      "download_gb": 15.8,
      "params_b": 30,
      "active_params_b": 3,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "reasons about code",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 2,
        "tokens_per_sec": 5.36,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by tencent",
        "ref": "hf.co/tencent/Hy-MT2-30B-A3B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 8.3129,
        "cost_breakdown": {
          "tokens_per_sec": 5.36,
          "usd_per_hour": 0.1605,
          "usd_per_million": 8.3129,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 5.5954
            },
            {
              "resource": "Memory",
              "quantity": "17.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 14.153,
              "usd_per_million": 2.6397
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0778
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "RTFgNEfiOHEhYFohoGoxFFyKell+w3kWlYqqiVggODJk9UopRoPwCwPSPfOtJV2qp4j5pZnjeRgHnIcpU0LeCg==",
          "signing_payload_b64": "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",
          "payload_sha256": "MntXlcKKYxYiFJcxaBqL2uG+JQGUu5LAZo85uJSjD3Y="
        },
        "build_page": "/inferences/hy-mt-2/30b-a3b/tencent-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/tencent/hy-mt2-30b-a3b-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b",
          "e3654f"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 2,
        "tokens_per_sec": 5.36,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by tencent",
        "ref": "hf.co/tencent/Hy-MT2-30B-A3B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 8.3129,
        "cost_breakdown": {
          "tokens_per_sec": 5.36,
          "usd_per_hour": 0.1605,
          "usd_per_million": 8.3129,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 5.5954
            },
            {
              "resource": "Memory",
              "quantity": "17.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 14.153,
              "usd_per_million": 2.6397
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0778
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "RTFgNEfiOHEhYFohoGoxFFyKell+w3kWlYqqiVggODJk9UopRoPwCwPSPfOtJV2qp4j5pZnjeRgHnIcpU0LeCg==",
          "signing_payload_b64": "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",
          "payload_sha256": "MntXlcKKYxYiFJcxaBqL2uG+JQGUu5LAZo85uJSjD3Y="
        },
        "build_page": "/inferences/hy-mt-2/30b-a3b/tencent-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/tencent/hy-mt2-30b-a3b-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b",
          "e3654f"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 2,
          "tokens_per_sec": 5.36,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by tencent",
          "ref": "hf.co/tencent/Hy-MT2-30B-A3B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 8.3129,
          "cost_breakdown": {
            "tokens_per_sec": 5.36,
            "usd_per_hour": 0.1605,
            "usd_per_million": 8.3129,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 5.5954
              },
              {
                "resource": "Memory",
                "quantity": "17.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 14.153,
                "usd_per_million": 2.6397
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0778
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "RTFgNEfiOHEhYFohoGoxFFyKell+w3kWlYqqiVggODJk9UopRoPwCwPSPfOtJV2qp4j5pZnjeRgHnIcpU0LeCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "MntXlcKKYxYiFJcxaBqL2uG+JQGUu5LAZo85uJSjD3Y="
          },
          "build_page": "/inferences/hy-mt-2/30b-a3b/tencent-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/tencent/hy-mt2-30b-a3b-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b",
            "e3654f"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "tencent/hy-mt2-30b-a3b",
        "url": "https://openrouter.ai/tencent/hy-mt2-30b-a3b",
        "input_usd_per_million": 0.074,
        "output_usd_per_million": 0.295,
        "free_tier": false,
        "context_tokens": 8192,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Tencent",
            "tag": "tencent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.074,
            "usd_per_m_out": 0.295,
            "context_length": 8192,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/tencent/Hy-MT2-30B-A3B"
      },
      "builds": 3,
      "variants": [
        {
          "model": "AX-Hy-MT2-30B-A3B-MLX-AXQ-MXFP4",
          "made_by": "AutomatosX",
          "page": "/inferences/hy-mt-2/30b-a3b~automatosx-ax-axq",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "5.4 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. About $8.31 per million tokens here, against OpenRouter's $0.29: the API is 28× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "phi-4-quantized.w8a8",
      "made_by": "redhatai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/60466e4b4f40b01b66151416/cdABRow21BL0sl1vSVTPk.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Text & chat",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/RedHatAI/phi-4-quantized.w8a8",
      "download_gb": null,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": 32768,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "reasons about code",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Marathi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": 32768,
        "build": null,
        "ref": "RedHatAI/phi-4-quantized.w8a8",
        "engine": "vllm",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-23T21:03:11.717Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/phi-4/base/redhatai-w8a8-vllm",
        "hub_url": "https://huggingface.co/redhatai/phi-4-quantized.w8a8",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 32768,
          "build": null,
          "ref": "RedHatAI/phi-4-quantized.w8a8",
          "engine": "vllm",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T21:03:11.717Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/base/redhatai-w8a8-vllm",
          "hub_url": "https://huggingface.co/redhatai/phi-4-quantized.w8a8",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded."
    },
    {
      "model": "Qwen2.5 0.5B Instruct",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-2-5/0.5b",
      "hub_url": "https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct",
      "download_gb": 0.5,
      "params_b": 0.5,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 32768,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "returns valid JSON",
        "long answers"
      ],
      "languages_passed": [
        "English",
        "Kannada",
        "Dutch",
        "Russian",
        "Sinhala"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux PC with an NVIDIA GPU, 64 GB",
        "machine_detail": null,
        "machine_class": "linux/amd64/nvidia/64g",
        "machines": 1,
        "tokens_per_sec": 319.04,
        "context_tokens": 32768,
        "build": "Q4_K_M (4-bit) by Qwen",
        "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:q4_k_m",
        "engine": "ollama",
        "cost_usd_per_million": 0.3508,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates",
        "measured_at": "2026-11-06T08:15:29.210Z",
        "signed": true,
        "build_page": null,
        "hub_url": "https://huggingface.co/qwen/qwen2.5-0.5b-instruct-gguf",
        "weeks": [
          "2026-W45"
        ]
      },
      "cheapest": {
        "machine": "Linux PC with an NVIDIA GPU, 32 GB",
        "machine_detail": null,
        "machine_class": "linux/amd64/nvidia/32g",
        "machines": 1,
        "tokens_per_sec": 239.08,
        "context_tokens": 32768,
        "build": "Q4_K_M (4-bit) by Qwen",
        "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:q4_k_m",
        "engine": "ollama",
        "cost_usd_per_million": 0.1288,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates",
        "measured_at": "2026-10-19T22:01:17.067Z",
        "signed": true,
        "build_page": null,
        "hub_url": "https://huggingface.co/qwen/qwen2.5-0.5b-instruct-gguf",
        "weeks": [
          "2026-W43"
        ]
      },
      "machine_kinds": 11,
      "machines_total": 12,
      "weeks_measured": [
        "2026-W31",
        "2026-W37",
        "2026-W38",
        "2026-W39",
        "2026-W40",
        "2026-W41",
        "2026-W42",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux PC with an NVIDIA GPU, 64 GB",
          "machine_detail": null,
          "machine_class": "linux/amd64/nvidia/64g",
          "machines": 1,
          "tokens_per_sec": 319.04,
          "context_tokens": 32768,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:q4_k_m",
          "engine": "ollama",
          "cost_usd_per_million": 0.3508,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates",
          "measured_at": "2026-11-06T08:15:29.210Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/qwen/qwen2.5-0.5b-instruct-gguf",
          "weeks": [
            "2026-W45"
          ]
        },
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": 247.91,
          "context_tokens": 8192,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 1.0352,
          "cost_breakdown": {
            "tokens_per_sec": 247.91,
            "usd_per_hour": 0.9239,
            "usd_per_million": 1.0352,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.6051
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.381,
                "usd_per_million": 0.0015
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.4034
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0252
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "zHgYIgtg5QgGhL3uf7V0VLRa/7/tlXLVJEZanHeu2Gwz89AeOtWygupmS8NxZVNrrm8hZsz70TkU4FIWXe9+Cg==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MjQ3LjkwNjc3MDEyNDYyOCwidW5pdCI6InRva2Vucy9zIiwidHMiOjE3OTQzMDk2NTI5OTUwMzIxMDAsIm5vZGVfZGlkIjoiZGlkOmVwbjowMDI0MDgwMTEyMjBiYzY1MGI1OWFmODc5NDg5NjAzMWQyZTgyYmNhMGUxMTlmMzkwZTIxZDY5MDk2M2FlNDc1NzgyMzgwODRiZjU0IiwiZXh0cmEiOnsiaGFyZHdhcmVfY2xhc3MiOiJ3aW5kb3dzL2FtZDY0L252aWRpYS8xNmciLCJtb2RlbCI6ImhmLmNvL1F3ZW4vUXdlbjIuNS0wLjVCLUluc3RydWN0LUdHVUY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hIiwic2FtcGxlX2NvdW50IjoxNDcsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6NjMuMjYxNjg0OTk2LCJ0b3RhbF90b2tlbnMiOjE1NjgzfX0=",
            "payload_sha256": "tNTIseiRt0C5b9dIx68HC99rXoydMVQFXz7SN7cITcE="
          },
          "build_page": "/inferences/qwen-2-5/0.5b/qwen-q4_k_m-ollama-148138f0",
          "hub_url": "https://huggingface.co/qwen/qwen2.5-0.5b-instruct-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows PC with an NVIDIA GPU, 16 GB",
          "machine_detail": null,
          "machine_class": "windows/amd64/nvidia/16g",
          "machines": 1,
          "tokens_per_sec": 247.91,
          "context_tokens": 32768,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:q4_k_m",
          "engine": "ollama",
          "cost_usd_per_million": 1.0352,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates",
          "measured_at": "2026-11-08T13:24:21.415Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/qwen/qwen2.5-0.5b-instruct-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W45"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 239.08,
          "context_tokens": 32768,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.5645,
          "cost_breakdown": {
            "tokens_per_sec": 239.08,
            "usd_per_hour": 0.4859,
            "usd_per_million": 0.5645,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.1255
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.381,
                "usd_per_million": 0.0016
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.4183
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0192
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:03:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "pxVUn4UHtls7r7NfLMqDWbFXQTwsgW6WdNzz8mQKHskLrOx7SQqyjnO4j7CUcB4hSAeqJTh+WKugSIQ9lSkBBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "digHmAeSiq2yn9kBZ2Tid4dg+mWV1kevdfYhcsGVr8w="
          },
          "build_page": "/inferences/qwen-2-5/0.5b/qwen-q4_k_m-ollama-148138f0",
          "hub_url": "https://huggingface.co/qwen/qwen2.5-0.5b-instruct-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux PC with an NVIDIA GPU, 32 GB",
          "machine_detail": null,
          "machine_class": "linux/amd64/nvidia/32g",
          "machines": 1,
          "tokens_per_sec": 239.08,
          "context_tokens": 32768,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:q4_k_m",
          "engine": "ollama",
          "cost_usd_per_million": 0.1288,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates",
          "measured_at": "2026-10-19T22:01:17.067Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/qwen/qwen2.5-0.5b-instruct-gguf",
          "weeks": [
            "2026-W43"
          ]
        },
        {
          "machine": "Windows PC with Intel graphics, 16 GB",
          "machine_detail": null,
          "machine_class": "windows/amd64/intel/16g",
          "machines": 2,
          "tokens_per_sec": 39.65,
          "context_tokens": 16384,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:q4_k_m",
          "engine": "ollama",
          "cost_usd_per_million": 1.7996,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates",
          "measured_at": "2026-11-14T10:38:26.168Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/qwen/qwen2.5-0.5b-instruct-gguf",
          "weeks": [
            "2026-W37",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ]
        },
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": 25.57,
          "context_tokens": 8192,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 2.791,
          "cost_breakdown": {
            "tokens_per_sec": 25.57,
            "usd_per_hour": 0.2569,
            "usd_per_million": 2.791,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 2.738
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.381,
                "usd_per_million": 0.0149
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.038
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "a8iCpfgwDpzHf+Mqc4d8mPd4RS4tj8MYRTmwI25HVFD0EdHCGjLAWDGRARoC5kJ7qhP0FCr3rDcyLeW7K2sFDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "rynD1JdS2lmJ50D8LUxRMmTYnUV7BCgc7PhDx7IDmeM="
          },
          "build_page": "/inferences/qwen-2-5/0.5b/qwen-q4_k_m-ollama-148138f0",
          "hub_url": "https://huggingface.co/qwen/qwen2.5-0.5b-instruct-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 18.98,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.6229,
          "cost_breakdown": {
            "tokens_per_sec": 18.98,
            "usd_per_hour": 0.1109,
            "usd_per_million": 1.6229,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.5809
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.381,
                "usd_per_million": 0.0201
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.022
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "D3oUPZtVj7YvoxG5YoLjPtEwJxK9MW0KQix7k9HPcOamJDUv1P/HriImF8RVo9Nfg01X//eWa1uFJk1UhRhiAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "fNGm9V5lYJl9qjzmctRr7iCus7qGJi5uOlSp26dxKyo="
          },
          "build_page": "/inferences/qwen-2-5/0.5b/qwen-q4_k_m-ollama-148138f0",
          "hub_url": "https://huggingface.co/qwen/qwen2.5-0.5b-instruct-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux PC with no GPU, 32 GB",
          "machine_detail": null,
          "machine_class": "linux/amd64/none/32g",
          "machines": 1,
          "tokens_per_sec": 18.98,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:q4_k_m",
          "engine": "ollama",
          "cost_usd_per_million": 1.6226,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates",
          "measured_at": "2026-10-26T05:13:45.047Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/qwen/qwen2.5-0.5b-instruct-gguf",
          "weeks": [
            "2026-W44"
          ]
        },
        {
          "machine": "Windows · NVIDIA 4 GB · 12 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 4 GB · 12 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 12,
            "cpu_model": null,
            "ram_gb": 7,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 4,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|12|7|nvidia|4|||"
          },
          "machine_class": "windows|amd64|12|7|nvidia|4|||",
          "machines": 1,
          "tokens_per_sec": 5.75,
          "context_tokens": 32768,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 37.5623,
          "cost_breakdown": {
            "tokens_per_sec": 5.75,
            "usd_per_hour": 0.7779,
            "usd_per_million": 37.5623,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 19.1223
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.381,
                "usd_per_million": 0.0662
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 17.3839
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.9898
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-22T10:54:02.459Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220b3960d56a82de98ec563fb48bbfc3ca6b4dc788b19545f4315f8f363aa9a1db0",
            "signature": "bTSzZMkpKaUGtatlxpE3nu4sGTK+PHYPBYUiLpW5XtP3Bytz5P2LvvvYvWYSprOWRAcjUL31UuqSTwSjbQaaBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "/KiVhGDZJmdErAW8JWT09zN/zcJrgeJ181/HdZQJENg="
          },
          "build_page": "/inferences/qwen-2-5/0.5b/qwen-q4_k_m-ollama-148138f0",
          "hub_url": "https://huggingface.co/qwen/qwen2.5-0.5b-instruct-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9a1db0"
          ]
        },
        {
          "machine": "Windows PC with an NVIDIA GPU, 8 GB",
          "machine_detail": null,
          "machine_class": "windows/amd64/nvidia/8g",
          "machines": 1,
          "tokens_per_sec": 5.75,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:q4_k_m",
          "engine": "ollama",
          "cost_usd_per_million": 25.2353,
          "cost_is_estimate": true,
          "cost_basis": "the network's metered rates, machine size assumed",
          "measured_at": "2026-09-16T09:37:16.913Z",
          "signed": true,
          "build_page": null,
          "hub_url": "https://huggingface.co/qwen/qwen2.5-0.5b-instruct-gguf",
          "weeks": [
            "2026-W38"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen2.5-0.5B-Instruct"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "319 tokens a second on a Linux PC with an NVIDIA GPU, 64 GB. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Gemma 3 270M",
      "made_by": "google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/gemma-3/270m",
      "hub_url": "https://huggingface.co/google/gemma-3-270m-it",
      "download_gb": 0.2,
      "params_b": 0.27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 32768,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "reasons about code",
        "returns valid JSON",
        "long answers"
      ],
      "languages_passed": [
        "English",
        "Bengali",
        "Tamil"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 291.82,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.7398,
        "cost_breakdown": {
          "tokens_per_sec": 291.82,
          "usd_per_hour": 0.7772,
          "usd_per_million": 0.7398,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 0.3769
            },
            {
              "resource": "Memory",
              "quantity": "0.2 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.196,
              "usd_per_million": 0.0007
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.3427
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0195
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "Zpk8DwNmxZpWQAp8UvvR1cs3ESn7OOqBA9uvazjfGikH8GDXupDYQQVssZmtHBPLyhbjtrCDSWbaitPrrebODw==",
          "signing_payload_b64": "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",
          "payload_sha256": "bVI8hPE20wRzMSYiO3EsoHBaoPDwzU6QOO4AUr9wyrQ="
        },
        "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
        "weeks": [
          "2026-W43",
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 291.82,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.7398,
        "cost_breakdown": {
          "tokens_per_sec": 291.82,
          "usd_per_hour": 0.7772,
          "usd_per_million": 0.7398,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 0.3769
            },
            {
              "resource": "Memory",
              "quantity": "0.2 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.196,
              "usd_per_million": 0.0007
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.3427
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0195
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "Zpk8DwNmxZpWQAp8UvvR1cs3ESn7OOqBA9uvazjfGikH8GDXupDYQQVssZmtHBPLyhbjtrCDSWbaitPrrebODw==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MjkxLjgyMTUzNDQyNDM1OTg3LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5NDAyODE5Mjk3MDk3MjIwNSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDUyNDA1MGU5Mjk2ZDMxMmNkMjgyOGM0MjYxNjViZmYyM2JhZWVmNzY5YWRhOTcwODMwOTg4MzJmMDZkYWU2NzkiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L252aWRpYS82NGciLCJtb2RlbCI6ImhmLmNvL3Vuc2xvdGgvZ2VtbWEtMy0yNzBtLWl0LUdHVUY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjUzLCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjkuODc1ODk4OTk5OTk5OTk1LCJ0b3RhbF90b2tlbnMiOjI4ODJ9fQ==",
          "payload_sha256": "bVI8hPE20wRzMSYiO3EsoHBaoPDwzU6QOO4AUr9wyrQ="
        },
        "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
        "weeks": [
          "2026-W43",
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 6,
      "machines_total": 15,
      "weeks_measured": [
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 291.82,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.7398,
          "cost_breakdown": {
            "tokens_per_sec": 291.82,
            "usd_per_hour": 0.7772,
            "usd_per_million": 0.7398,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 0.3769
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.0007
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.3427
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0195
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "Zpk8DwNmxZpWQAp8UvvR1cs3ESn7OOqBA9uvazjfGikH8GDXupDYQQVssZmtHBPLyhbjtrCDSWbaitPrrebODw==",
            "signing_payload_b64": "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",
            "payload_sha256": "bVI8hPE20wRzMSYiO3EsoHBaoPDwzU6QOO4AUr9wyrQ="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": 286.08,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 0.8964,
          "cost_breakdown": {
            "tokens_per_sec": 286.08,
            "usd_per_hour": 0.9232,
            "usd_per_million": 0.8964,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.5243
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.0007
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.3495
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0218
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "gHIs9srClOIWui6TB0a8u3+HYoQ2InpMQsoIQ1ZePCmEE6rV/FV0eksjOcgLUbFGt48dsyxNJ87TFrKN7/PEAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "Mo4jh+tDGOYKA4gL1wH8vXyFhpK+oQvtLnWGcnukMQs="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 68.95,
          "context_tokens": null,
          "build": "IQ2_M (2-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_M",
          "engine": "ollama",
          "cost_usd_per_million": 1.0314,
          "cost_breakdown": {
            "tokens_per_sec": 68.95,
            "usd_per_hour": 0.256,
            "usd_per_million": 1.0314,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 1.0152
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.142,
                "usd_per_million": 0.0021
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0141
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "9G+CFgOxofbjtvP7XJaOsiscKDpQv5maZUbkMdsMYyApDt7kF7YpXIjzIV9V+PiiTUBP7sUL7DaaUjtOYPlVAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "HmvwBeRg3+rEpGBWlrg50P1+q+7bt6atl5pj0nUNLFo="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-iq2_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 52.33,
          "context_tokens": null,
          "build": "IQ3_XXS (3-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ3_XXS",
          "engine": "ollama",
          "cost_usd_per_million": 1.3589,
          "cost_breakdown": {
            "tokens_per_sec": 52.33,
            "usd_per_hour": 0.256,
            "usd_per_million": 1.3589,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 1.3376
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.142,
                "usd_per_million": 0.0027
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0186
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "/1h1wIBUh4G0+vn8FkW/uITxL+O0cvTgB/uH5SVX+DRWG051m/qlM9dnwIGRBXfpbLlZjxWlxEwMMAamp2plCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "5Buy1YIHoXmtncx3nB+f5xn/+OBc9XyDTkPHB5N5kjE="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": 51.51,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 1.3816,
          "cost_breakdown": {
            "tokens_per_sec": 51.51,
            "usd_per_hour": 0.2562,
            "usd_per_million": 1.3816,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 1.359
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.0038
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0189
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "zkcES8cXZ7skzNxSnPib3/T8sm2V2+s6cNUllSKND5IaWSR+SfJjOG/AmbMbCZoTWpiwkoiEzDy5W5zNbnwzAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "3Mw+WsNzZKxl/2a2mJc2kHTEXsWOP/Q4ICa38Jn9DBQ="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 41.45,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 1.717,
          "cost_breakdown": {
            "tokens_per_sec": 41.45,
            "usd_per_hour": 0.2562,
            "usd_per_million": 1.717,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 1.6888
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.0047
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0235
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "LogKNw4iHnjAhtX6trv+yW4zW50T6yMOsxPBRmTg2WFXJYo7JjXJM05a5M9D5COC6kwvcHaY1vBFa6sKx5qyBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "hnv3vpJ0FPXAT/elYqji8cFZffMzzhL26Rkpbp7BchA="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 37.38,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.819,
          "cost_breakdown": {
            "tokens_per_sec": 37.38,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.819,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8026
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.0052
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0112
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "4Hd1CZ/Lu7kOHzU4Y01dgGPHZBY/pVn0K9rz71a+qYin2redirsNjdTJGProHQgzGttRM9dScpjWJeuBwiH1DA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MzcuMzc4NzEyMDQwODYzMjEsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkyOTE0MDc0NjQwODkyNjI3LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwMjY3ZWI1Zjg5ZDg4YmZlZDFmYjMzNzk4Yzk2ZDZlYTk2Mzk5ZjM5MGNjNjgxNWJhZWM3ZDljYjE3N2FjYmU3NSIsImV4dHJhIjp7Im1vZGVsIjoiaGYuY28vdW5zbG90aC9nZW1tYS0zLTI3MG0taXQtR0dVRjpRNF9LX00iLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6NDUsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6OTguMTMwNzIxOTk5OTk5OTksInRvdGFsX3Rva2VucyI6MzY2OH19",
            "payload_sha256": "UdbAtK7zYHWB3yzf4ksBncfsrzxy5AFfRdAV0TxxqOs="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 34.6,
          "context_tokens": null,
          "build": "IQ2_XXS (2-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_XXS",
          "engine": "ollama",
          "cost_usd_per_million": 2.0551,
          "cost_breakdown": {
            "tokens_per_sec": 34.6,
            "usd_per_hour": 0.256,
            "usd_per_million": 2.0551,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 2.023
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.139,
                "usd_per_million": 0.004
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0281
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "/Oz4CzLFCswmhw6znYJCNBw7jG7Au4+JDS64ADkWcjOStvTXT1WpuDpUGIWcELyQdZ8tInEPi626Wh88hlglAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "DgCqjlr+cP8mdeFOqh2MkfP0G0OSNhBNJn/OSfh2z6E="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-iq2_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 34.58,
          "context_tokens": null,
          "build": "IQ2_M (2-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.8836,
          "cost_breakdown": {
            "tokens_per_sec": 34.58,
            "usd_per_hour": 0.11,
            "usd_per_million": 0.8836,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8674
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.142,
                "usd_per_million": 0.0041
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0121
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "Z9INea8tByqLpWMDkSTpSec2Tpr97QRKv2kVQENcIsDN8FFpyYLQmZ99akLda7ya/1mN/eQYkhaUSNHNF/U2Ag==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MzQuNTg0Mjg0MjAxNzE4NTcsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkyOTE0MDc0NjQwODkyNjI3LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwMjY3ZWI1Zjg5ZDg4YmZlZDFmYjMzNzk4Yzk2ZDZlYTk2Mzk5ZjM5MGNjNjgxNWJhZWM3ZDljYjE3N2FjYmU3NSIsImV4dHJhIjp7Im1vZGVsIjoiaGYuY28vdW5zbG90aC9nZW1tYS0zLTI3MG0taXQtR0dVRjpJUTJfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo0Nywic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjoxNjIuOTY0MTk0MDAwMDAwMDIsInRvdGFsX3Rva2VucyI6NTYzNn19",
            "payload_sha256": "lS1Yf+wdEKPa7A7s34G7ywsOwgCAYVk5Hqtq6n8QRRI="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-iq2_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W46"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 34.48,
          "context_tokens": null,
          "build": "IQ2_XXS (2-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_XXS",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.8863,
          "cost_breakdown": {
            "tokens_per_sec": 34.48,
            "usd_per_hour": 0.11,
            "usd_per_million": 0.8863,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8702
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.139,
                "usd_per_million": 0.004
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0121
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "nln3fof0ALK0dGfVfn4TNpH4sH153ckDAnIxAQ3iWKmKFDENXKemnOt6AGtf7YRtChTkzVFGM2xrupYIM5lgCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "4c7bbWMNMH/Ned5wuRTIV5YqCtNJKL0lhA2f338/SsA="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-iq2_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W46"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 33.41,
          "context_tokens": null,
          "build": "IQ3_XXS (3-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9147,
          "cost_breakdown": {
            "tokens_per_sec": 33.41,
            "usd_per_hour": 0.11,
            "usd_per_million": 0.9147,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8979
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.142,
                "usd_per_million": 0.0043
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0125
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "uL2ZwCfnlyAc/7vRh+ojNNuVAZORMQu5PJH12WYmdZSYqyKhz2V0t2q97999UmEsz+Va75g+XiPh7YBy06MIDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "S2vxWz9LesteHVhdc2E1D43oCbOsgkODdRVgbAd3OxA="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W46"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 33,
          "context_tokens": null,
          "build": "Q4_0 (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9274,
          "cost_breakdown": {
            "tokens_per_sec": 33,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.9274,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.9091
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.187,
                "usd_per_million": 0.0057
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0126
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "FTZFv80vKJMZD1WcVJ8ofu+ljRDHev7FLjlP4QdwyMTpd+EDAhhRv4XpejKcTq+1JC/jCfknhfbwRkA6wzQZCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "fAYSM1jXIZt26pxWTxa0u+A4RmkO3qtduPPIW6v151w="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 30.86,
          "context_tokens": 32768,
          "build": "Q8_0 (8-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.993,
          "cost_breakdown": {
            "tokens_per_sec": 30.86,
            "usd_per_hour": 0.1103,
            "usd_per_million": 0.993,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.9722
              },
              {
                "resource": "Memory",
                "quantity": "0.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.226,
                "usd_per_million": 0.0073
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0135
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "lIWNsN60xFm9JhtL4leDzKdAYN9HzbDvdleNcJvhqMI3IV//XCwCLtmFz4G3mzmtOiZA90dk+UcWxMdln/TTCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "ItRcKZ631zh9KtJ2KdbLchUokU/bSpj+V7CHn+Le3GM="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-q8_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.22631",
            "cpus": 8,
            "cpu_model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
            "ram_gb": 7,
            "gpu_vendor": "AMD",
            "gpu_model": "AMD Radeon(TM) Vega 8 Mobile Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "amd",
            "host_org": null,
            "key": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 27.34,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 6.412,
          "cost_breakdown": {
            "tokens_per_sec": 27.34,
            "usd_per_hour": 0.6312,
            "usd_per_million": 6.412,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 2.5599
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.0072
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 3.657
              },
              {
                "resource": "Energy",
                "quantity": "185 W modelled (7 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.139,
                "usd_per_million": 0.1879
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 997.6,
              "multi_core_hashes_per_sec": 8516.9,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T08:14:32.559Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "memory": {
              "total_gb": 7,
              "bandwidth_gb_s": 6,
              "latency_ns": 61.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T08:14:33.251Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2347.3,
              "write_mb_s": 204.2,
              "random_read_iops": 96070,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-10-29T08:14:33.869Z",
              "measured_in": "IN_560001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-29T09:04:03.140Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220c04e56f30abc1d9b2ab523a1f562860a830eec9a1aeb060ac69cd1d1ded061c9",
            "signature": "GLnSBD5ynTuGblHtKb33m4ubr9VqLhS0BBv1ntoUsasiXw0hjsgDlonfSlamGyX7e06BtkJE3TXmajlN/h8vBw==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MjcuMzQ0NjQ3ODg2MzQ4ODQsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzMjY0NjQzMTQwMzQ2OTAwLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwYzA0ZTU2ZjMwYWJjMWQ5YjJhYjUyM2ExZjU2Mjg2MGE4MzBlZWM5YTFhZWIwNjBhYzY5Y2QxZDFkZWQwNjFjOSIsImV4dHJhIjp7Im1vZGVsIjoiaGYuY28vdW5zbG90aC9nZW1tYS0zLTI3MG0taXQtR0dVRjpRNF9LX00iLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6NDYsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6NDguODIxMjU0LCJ0b3RhbF90b2tlbnMiOjEzMzV9fQ==",
            "payload_sha256": "i7cif1bMTlpocFTMyueR/+vOcEjh9BAhv1dDjBazT7A="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "d061c9"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 23.52,
          "context_tokens": 16384,
          "build": "F16 (16-bit, full precision) by unsloth",
          "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:F16",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.3114,
          "cost_breakdown": {
            "tokens_per_sec": 23.52,
            "usd_per_hour": 0.111,
            "usd_per_million": 1.3114,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.2757
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.421,
                "usd_per_million": 0.0179
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0177
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "L6sfn2qL4akIScivTrr6c7xREiH+1qjCgpqGkwqOoQEmCs+yYs6zSzpdCJPb7nX5rgb12IlEez5EKSuvSbPUBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "rSJP825Ft1QwPah6P9ou8Avoogn0rR+U1gA3RD0T3RE="
          },
          "build_page": "/inferences/gemma-3/270m/unsloth-f16-ollama",
          "hub_url": "https://huggingface.co/unsloth/gemma-3-270m-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Gemma terms of use",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/google/gemma-3-270m-it"
      },
      "builds": 7,
      "variants": [],
      "bottom_line": "292 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Phi 4 Mini",
      "made_by": "microsoft",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1583646260758-5e64858c87403103f9f1055d.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/phi-4/mini",
      "hub_url": "https://huggingface.co/microsoft/Phi-4-mini-instruct",
      "download_gb": 2.3,
      "params_b": 4.1,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "English",
        "Bengali",
        "Hindi",
        "Dutch",
        "Russian",
        "Malayalam",
        "Tamil"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": "AMD Ryzen 7 5800X 8-Core",
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "GeForce RTX 3080 Ti",
          "gpu_count": 1,
          "vram_gb": 12,
          "host": "own machine",
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
        "machines": 1,
        "tokens_per_sec": 205.49,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by maziyarpanahi",
        "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": 1.247,
        "cost_breakdown": {
          "tokens_per_sec": 205.49,
          "usd_per_hour": 0.9225,
          "usd_per_million": 1.247,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "15 of 16 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 150,
              "usd_per_million": 0.73
            },
            {
              "resource": "Memory",
              "quantity": "0.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0,
              "usd_per_million": 0
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.4866
            },
            {
              "resource": "Energy",
              "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 6.25,
              "usd_per_million": 0.0304
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-10T11:20:52.995Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "TYEeuh3rrSdgvu/LsNQESePjsV7jbCoYVOsqpoJtatkcuOC1tx2UNb/U6u3bXevDGiqKWxEMMkYOUgpX9JI2DQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "FhWLGFN+H10MATforjUzWFeZUrrjRYW/xNW/m83asiQ="
        },
        "build_page": "/inferences/phi-4/mini/maziyarpanahi-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/maziyarpanahi/phi-4-mini-instruct-gguf",
        "weeks": [
          "2026-W38",
          "2026-W45"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "cheapest": {
        "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": "AMD Ryzen 7 5800X 8-Core",
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "GeForce RTX 3080 Ti",
          "gpu_count": 1,
          "vram_gb": 12,
          "host": "own machine",
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
        "machines": 1,
        "tokens_per_sec": 205.49,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by maziyarpanahi",
        "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": 1.247,
        "cost_breakdown": {
          "tokens_per_sec": 205.49,
          "usd_per_hour": 0.9225,
          "usd_per_million": 1.247,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "15 of 16 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 150,
              "usd_per_million": 0.73
            },
            {
              "resource": "Memory",
              "quantity": "0.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0,
              "usd_per_million": 0
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.4866
            },
            {
              "resource": "Energy",
              "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 6.25,
              "usd_per_million": 0.0304
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-10T11:20:52.995Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "TYEeuh3rrSdgvu/LsNQESePjsV7jbCoYVOsqpoJtatkcuOC1tx2UNb/U6u3bXevDGiqKWxEMMkYOUgpX9JI2DQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "FhWLGFN+H10MATforjUzWFeZUrrjRYW/xNW/m83asiQ="
        },
        "build_page": "/inferences/phi-4/mini/maziyarpanahi-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/maziyarpanahi/phi-4-mini-instruct-gguf",
        "weeks": [
          "2026-W38",
          "2026-W45"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "machine_kinds": 8,
      "machines_total": 27,
      "weeks_measured": [
        "2026-W35",
        "2026-W37",
        "2026-W38",
        "2026-W42",
        "2026-W43",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": 205.49,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 1.247,
          "cost_breakdown": {
            "tokens_per_sec": 205.49,
            "usd_per_hour": 0.9225,
            "usd_per_million": 1.247,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.73
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.4866
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0304
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "TYEeuh3rrSdgvu/LsNQESePjsV7jbCoYVOsqpoJtatkcuOC1tx2UNb/U6u3bXevDGiqKWxEMMkYOUgpX9JI2DQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "FhWLGFN+H10MATforjUzWFeZUrrjRYW/xNW/m83asiQ="
          },
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W38",
            "2026-W45"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 154.35,
          "context_tokens": null,
          "build": "Q2_K (2-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q2_K",
          "engine": "ollama",
          "cost_usd_per_million": 1.6602,
          "cost_breakdown": {
            "tokens_per_sec": 154.35,
            "usd_per_hour": 0.9225,
            "usd_per_million": 1.6602,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.9718
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.6479
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0405
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "KOyxMTtxouNhSfmC7bDu0EUmNL9GQ9iazowJLfugrEgYIJy3Ql2y9i2QAlrnhnzGReyg71yTThQnhqk1KA83Bg==",
            "signing_payload_b64": "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",
            "payload_sha256": "b+ywwgvAtpfwIzpnQPj2j8BHBHtcFE/Z7RHHDrs2OLE="
          },
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q2_k-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W38"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 134.8,
          "context_tokens": 4096,
          "build": "Q3_K_S (3-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_S",
          "engine": "ollama",
          "cost_usd_per_million": 1.901,
          "cost_breakdown": {
            "tokens_per_sec": 134.8,
            "usd_per_hour": 0.9225,
            "usd_per_million": 1.901,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 1.1128
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.7419
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0464
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "cYpdYJUDoS8tfO2l7N0LI9CEPLFy5oGszNDtIcbbqwcz1z/h2jLAyPH2u9r2aXley/nzSrkZbu8/bi5vNVVaDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "dBN74GKPQzjce4PDsuzeLXznH7ImC9MNAFVOlSIw30k="
          },
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W38"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 110.09,
          "context_tokens": 131072,
          "build": "Q3_K_L (3-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_L",
          "engine": "ollama",
          "cost_usd_per_million": 2.3277,
          "cost_breakdown": {
            "tokens_per_sec": 110.09,
            "usd_per_hour": 0.9225,
            "usd_per_million": 2.3277,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 1.3625
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.9084
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0568
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "7ylTDBueNvVUiJZL2O8MUujv+SpZp4bhF1QWaSIVqSjw0kP6Gk/xqeLuwNnmPF3EWL1F13cgHMnBdOn2FZPQAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "CHRwQb/D4ofELxYWOxKBsE1kJyJThAgerSzxGGZEPjA="
          },
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q3_k_l-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 79.36,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.7424,
          "cost_breakdown": {
            "tokens_per_sec": 79.36,
            "usd_per_hour": 0.7835,
            "usd_per_million": 2.7424,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 1.3862
              },
              {
                "resource": "Memory",
                "quantity": "2.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.934,
                "usd_per_million": 0.0244
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.2601
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0718
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "ZVrH7Uw3fgyUscrLT6EXcr8fmS3oA1+N2BRHjNs/ka2uCbqy6FNUJEtXa3kzJaFVKwDxZREZzhGcBqx97OMsCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "6XmtTI7MMrj3CZInM9+gejn4BOVgLv/ws5x5QXv+o+U="
          },
          "build_page": "/inferences/phi-4/mini/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 77.39,
          "context_tokens": null,
          "build": "Q4_K_S (4-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q4_K_S",
          "engine": "ollama",
          "cost_usd_per_million": 3.3111,
          "cost_breakdown": {
            "tokens_per_sec": 77.39,
            "usd_per_hour": 0.9225,
            "usd_per_million": 3.3111,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 1.9382
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.2921
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0808
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "t2qkAy0C+sPNIaON6cHhKeEnjsLTaGPE9z2zw2jVUkN/32v8vxPx1OzeKT9iOOv18fFV2cV8Srb1HWgWMEDxAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "DiQp5zspuoHVy+qH2LoVx2ntYmh68NMEDHIsXcwF5Dw="
          },
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q4_k_s-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W38"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 74.2,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.8399,
          "cost_breakdown": {
            "tokens_per_sec": 74.2,
            "usd_per_hour": 0.4915,
            "usd_per_million": 1.8399,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.4043
              },
              {
                "resource": "Memory",
                "quantity": "2.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.934,
                "usd_per_million": 0.0261
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.3477
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0618
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "kAbAwjkvdsBFOCbF1W8lfWET8F5ase9u7cUpkZyX+VlsI2O2767fj/bZL3UZ19P9g7wYf/hb8WyhITV69XjgBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "SEJrC5gK/PJphBAsMw/UpXXofHpoFPVGcKKiD6Il5mc="
          },
          "build_page": "/inferences/phi-4/mini/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 68.61,
          "context_tokens": null,
          "build": "Q2_K (2-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q2_K",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.9806,
          "cost_breakdown": {
            "tokens_per_sec": 68.61,
            "usd_per_hour": 0.4892,
            "usd_per_million": 1.9806,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.4373
              },
              {
                "resource": "Memory",
                "quantity": "1.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.305,
                "usd_per_million": 0.019
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.4575
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0668
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "miWlotwURDVdfj8A9vdzRloamTA7+fI14TpV23WXNkOwY0VVhzEmoVcdW8q/IcwwQTwhCD1NggetH9V/Nyl+Cw==",
            "signing_payload_b64": "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",
            "payload_sha256": "4STfVZIT9Rn2TEV8q8YwHKcoYfwObI5ixfYMYGM/nps="
          },
          "build_page": "/inferences/phi-4/mini/unsloth-q2_k-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 66.57,
          "context_tokens": 131072,
          "build": "Q3_K_M (3-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q3_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.0463,
          "cost_breakdown": {
            "tokens_per_sec": 66.57,
            "usd_per_hour": 0.4904,
            "usd_per_million": 2.0463,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.4506
              },
              {
                "resource": "Memory",
                "quantity": "2.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.643,
                "usd_per_million": 0.0247
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.5022
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0688
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "AKcFh//Njg8iVOysLcmxh9/yN2azllYq1iVbZqyTixUxcLAnN3gD4eXVa+FIK/xMlJV55iaE/L1aPYkXgq8XAA==",
            "signing_payload_b64": "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",
            "payload_sha256": "yY3qtZpEj7hF/3GYz7gsTXh+0QIMVSH4/+S67ncU0a4="
          },
          "build_page": "/inferences/phi-4/mini/unsloth-q3_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 64.06,
          "context_tokens": 131072,
          "build": "Q3_K_L (3-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.1009,
          "cost_breakdown": {
            "tokens_per_sec": 64.06,
            "usd_per_hour": 0.4845,
            "usd_per_million": 2.1009,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.4683
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.5611
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0715
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "ddtE+Hqpxa/EFv19wHPL6wEkyjCIdjPLgGqlhFJrEBONKtMLdOT1tf01fDEcnkrONVRcLk+OFdzsOiduSPziAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "6VIQ5W2keskTi1PiCyFG5CGsOlYlFS4FVIot+eTbBT8="
          },
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q3_k_l-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 39.21,
          "context_tokens": null,
          "build": "Q8_0 (8-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q8_0",
          "engine": "ollama",
          "cost_usd_per_million": 6.5351,
          "cost_breakdown": {
            "tokens_per_sec": 39.21,
            "usd_per_hour": 0.9225,
            "usd_per_million": 6.5351,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 3.8254
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.5503
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.1594
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "0/2WQviYqJwPbo3QvSX1sNfzcdaUjb4qQwthVAbyyAJW71qosWL3uULyYAPHSguF1FEmyyT2ATCPRpHMpza4AA==",
            "signing_payload_b64": "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",
            "payload_sha256": "A3Q9puLABfdtMMlZiXzlvOnfSQvJQ2xUQGFbYgZnO3I="
          },
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q8_0-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W38"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 36.2,
          "context_tokens": 4096,
          "build": "IQ4_XS (4-bit) by bartowski",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ4_XS",
          "engine": "ollama-host",
          "cost_usd_per_million": 5.4463,
          "cost_breakdown": {
            "tokens_per_sec": 36.2,
            "usd_per_hour": 0.7097,
            "usd_per_million": 5.4463,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 2.4864
              },
              {
                "resource": "Memory",
                "quantity": "2.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.726,
                "usd_per_million": 0.0477
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.7626
              },
              {
                "resource": "Energy",
                "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.417,
                "usd_per_million": 0.1497
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "hx1acBZQso+nYFg4CiuuSEh08F2QiviMzROVBjOoL/rl5IdfExrOt1CCHgrgyYQRhtMWwOQHY8IU4zdS6Q0OCA==",
            "signing_payload_b64": "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",
            "payload_sha256": "I4lIVgQ6kUWnZ01OoU16C2ieBRZz7hZkcbWuwyVMtyE="
          },
          "build_page": "/inferences/phi-4/mini/bartowski-iq4_xs-ollama",
          "hub_url": "https://huggingface.co/bartowski/microsoft_phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W42"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 33.74,
          "context_tokens": 4096,
          "build": "Q4_0 (4-bit) by bartowski",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 5.843,
          "cost_breakdown": {
            "tokens_per_sec": 33.74,
            "usd_per_hour": 0.7097,
            "usd_per_million": 5.843,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 2.6674
              },
              {
                "resource": "Memory",
                "quantity": "2.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.726,
                "usd_per_million": 0.0512
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.9638
              },
              {
                "resource": "Energy",
                "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.417,
                "usd_per_million": 0.1606
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "x7z9ZPKprNiQ3A1wl2wFWNVH9L5yDKBpr73E1l3ThovkhvbrLGND3jWAWd2sXRfWdG2dY5Nmvo9QZX0xQ4iVAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "xR29jlBezGlhrFJPRjm/sFgHjG9fMsg/7UPO0ublYIc="
          },
          "build_page": "/inferences/phi-4/mini/bartowski-q4_0-ollama",
          "hub_url": "https://huggingface.co/bartowski/microsoft_phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W42"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 33.7,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 5.8504,
          "cost_breakdown": {
            "tokens_per_sec": 33.7,
            "usd_per_hour": 0.7097,
            "usd_per_million": 5.8504,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 2.6709
              },
              {
                "resource": "Memory",
                "quantity": "2.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.726,
                "usd_per_million": 0.0512
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.9676
              },
              {
                "resource": "Energy",
                "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.417,
                "usd_per_million": 0.1608
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "krq4gBbUnikVohZgyuVVfAabzQ5JC2HIDrElFzyrPLcbexS+jBgrWnBxB+zZVVch0h+kibI3di57mYmMwrEQAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "NiRAaCmNiqGv77eYFDq/H4m9BDBsigLdPw+JhLnSTVo="
          },
          "build_page": "/inferences/phi-4/mini/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/microsoft_phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W42"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Mac · no GPU · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|none|0|||"
          },
          "machine_class": "darwin|arm64|10|16|none|0|||",
          "machines": 1,
          "tokens_per_sec": 28.18,
          "context_tokens": 8192,
          "build": "Q3_K_S (3-bit) by bartowski",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 3.2992,
          "cost_breakdown": {
            "tokens_per_sec": 28.18,
            "usd_per_hour": 0.3347,
            "usd_per_million": 3.2992,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 3.1936
              },
              {
                "resource": "Memory",
                "quantity": "2.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.726,
                "usd_per_million": 0.0612
              },
              {
                "resource": "Energy",
                "quantity": "45 W modelled (9 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 1.25,
                "usd_per_million": 0.0444
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "8Xz1GAWAASCaz8+14tIwqc5fiIo7WrZxPRbU6VuPKFVdKDzCEaDKp0AeJ0nE9LYguxZRietvAF3+sG9FEOg2Aw==",
            "signing_payload_b64": "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",
            "payload_sha256": "KHVOPerhOmuf3iqavx9O84HEDTDTO+tHz7sOHBbiJPY="
          },
          "build_page": "/inferences/phi-4/mini/bartowski-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/bartowski/microsoft_phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W35"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 27.09,
          "context_tokens": 4096,
          "build": "IQ2_M (2-bit) by bartowski",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ2_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 7.2781,
          "cost_breakdown": {
            "tokens_per_sec": 27.09,
            "usd_per_hour": 0.7097,
            "usd_per_million": 7.2781,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 3.3226
              },
              {
                "resource": "Memory",
                "quantity": "2.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.726,
                "usd_per_million": 0.0637
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 3.6918
              },
              {
                "resource": "Energy",
                "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.417,
                "usd_per_million": 0.2
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "5pnmmnCBFX7YVBu9HoguAmWP+UuHO/wHarFN2zDBCHYuXG2Ka10uYsIIcpuHLBpDFwR1hub6M868yNuzENy5Bg==",
            "signing_payload_b64": "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",
            "payload_sha256": "f5aMCnoE37Ik7MwAQNtQ2SvCk0GoQioSuiFw5jINQvA="
          },
          "build_page": "/inferences/phi-4/mini/bartowski-iq2_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/microsoft_phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W42"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "windows|amd64|8|16|none|0|||"
          },
          "machine_class": "windows|amd64|8|16|none|0|||",
          "machines": 1,
          "tokens_per_sec": 4.41,
          "context_tokens": null,
          "build": "IQ2_XS (2-bit) by mungert",
          "ref": "hf.co/Mungert/Phi-4-mini-instruct.gguf:IQ2_XS",
          "engine": "ollama",
          "cost_usd_per_million": 16.1008,
          "cost_breakdown": {
            "tokens_per_sec": 4.41,
            "usd_per_hour": 0.2555,
            "usd_per_million": 16.1008,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 15.8803
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.2205
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "oRPz1Ag39CANoFCV2iueHhXS+bHpN8bRBOpUlfiru6y9bY3EjNjPRX18+BkIVxPEfqKJ0Jvu+b333nJMwvFmBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "NwtWyN5ntWcpOP1iFxECpVTLIoSlYWtmnGeFS74tAlg="
          },
          "build_page": "/inferences/phi-4/mini/mungert-iq2_xs-ollama",
          "hub_url": "https://huggingface.co/mungert/phi-4-mini-instruct.gguf",
          "weeks": [
            "2026-W37"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Mac · no GPU · 4 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 4 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 12.7.6",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|amd64|4|8|none|0|||"
          },
          "machine_class": "darwin|amd64|4|8|none|0|||",
          "machines": 1,
          "tokens_per_sec": 2.2,
          "context_tokens": 131072,
          "build": "Q3_K_L (3-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "cost_usd_per_million": 13.851,
          "cost_breakdown": {
            "tokens_per_sec": 2.2,
            "usd_per_hour": 0.1095,
            "usd_per_million": 13.851,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 13.6611
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1899
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Core i5-5350U",
              "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
              "clock_ghz": 1.8,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 3,
              "single_core_hashes_per_sec": 93.7,
              "multi_core_hashes_per_sec": 195.1,
              "sustained_ratio": 0.7,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T23:57:35.534Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 8,
              "bandwidth_gb_s": 0.1,
              "latency_ns": 170.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T23:57:37.012Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 89.9,
              "write_mb_s": 44.3,
              "random_read_iops": 25045,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T23:57:40.536Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-16T00:27:19.106Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "mu9Xqlwj5QnUrGKvvBo4uSzWdN0mApaH848gtB9ZuhkOJybjp1OOg5ukwpKxns+TNs4N06CGGDRgwWPRbUIwCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "Dl9vP8+iQMF4OzZquCYd6meUIJ97lsBlH4HTtgNH3/4="
          },
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q3_k_l-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "d68672"
          ]
        },
        {
          "machine": "Mac · no GPU · 4 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 4 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 12.7.6",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|amd64|4|8|none|0|||"
          },
          "machine_class": "darwin|amd64|4|8|none|0|||",
          "machines": 1,
          "tokens_per_sec": 2.18,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 14.8637,
          "cost_breakdown": {
            "tokens_per_sec": 2.18,
            "usd_per_hour": 0.1165,
            "usd_per_million": 14.8637,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 13.7835
              },
              {
                "resource": "Memory",
                "quantity": "2.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.934,
                "usd_per_million": 0.8886
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1916
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Core i5-5350U",
              "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
              "clock_ghz": 1.8,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 3,
              "single_core_hashes_per_sec": 93.7,
              "multi_core_hashes_per_sec": 195.1,
              "sustained_ratio": 0.7,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T23:57:35.534Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 8,
              "bandwidth_gb_s": 0.1,
              "latency_ns": 170.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T23:57:37.012Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 89.9,
              "write_mb_s": 44.3,
              "random_read_iops": 25045,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T23:57:40.536Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-16T00:27:19.106Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "SD/2MwO6SAmRC9FzH+Yex+vRUEjvBzKtDzcmztrdEWXxI12t1DXhhwnGUnSPM+dcgXXOC6v8z3qgzJpY0b4JAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "vqOr+XhSD4QjGsTzXpbR3XwvKJNhmWa5uj2/g0QG8rs="
          },
          "build_page": "/inferences/phi-4/mini/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "d68672"
          ]
        },
        {
          "machine": "Mac · no GPU · 4 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 4 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 12.7.6",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|amd64|4|8|none|0|||"
          },
          "machine_class": "darwin|amd64|4|8|none|0|||",
          "machines": 1,
          "tokens_per_sec": 2.17,
          "context_tokens": 4096,
          "build": "Q3_K_S (3-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 14.0402,
          "cost_breakdown": {
            "tokens_per_sec": 2.17,
            "usd_per_hour": 0.1095,
            "usd_per_million": 14.0402,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 13.8477
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1925
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Core i5-5350U",
              "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
              "clock_ghz": 1.8,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 3,
              "single_core_hashes_per_sec": 93.7,
              "multi_core_hashes_per_sec": 195.1,
              "sustained_ratio": 0.7,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T23:57:35.534Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 8,
              "bandwidth_gb_s": 0.1,
              "latency_ns": 170.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T23:57:37.012Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 89.9,
              "write_mb_s": 44.3,
              "random_read_iops": 25045,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T23:57:40.536Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-16T00:27:19.106Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "im4DmT2oyHpzCWSA/HqOmUpmlDKFWe802H3/bs2cHY/g4kvzEXDbjeL06f+Xk7dK6J20wYgSFLHq5R7KhW28Dg==",
            "signing_payload_b64": "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",
            "payload_sha256": "e9ig+TjBpsOPX+7wK1J58zCA3krWI8QZfGOBd3zz29c="
          },
          "build_page": "/inferences/phi-4/mini/maziyarpanahi-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W38"
          ],
          "providers": [
            "d68672"
          ]
        },
        {
          "machine": "Mac · no GPU · 4 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 4 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 12.7.6",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|amd64|4|8|none|0|||"
          },
          "machine_class": "darwin|amd64|4|8|none|0|||",
          "machines": 1,
          "tokens_per_sec": 1.06,
          "context_tokens": 4096,
          "build": "IQ2_M (2-bit) by bartowski",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ2_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 30.3882,
          "cost_breakdown": {
            "tokens_per_sec": 1.06,
            "usd_per_hour": 0.1157,
            "usd_per_million": 30.3882,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 28.3622
              },
              {
                "resource": "Memory",
                "quantity": "2.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.726,
                "usd_per_million": 1.6318
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.3942
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Core i5-5350U",
              "model_raw": "Intel(R) Core(TM) i5-5350U CPU @ 1.80GHz",
              "clock_ghz": 1.8,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 3,
              "single_core_hashes_per_sec": 93.7,
              "multi_core_hashes_per_sec": 195.1,
              "sustained_ratio": 0.7,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T23:57:35.534Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 8,
              "bandwidth_gb_s": 0.1,
              "latency_ns": 170.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T23:57:37.012Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 89.9,
              "write_mb_s": 44.3,
              "random_read_iops": 25045,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T23:57:40.536Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-16T00:27:19.106Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220841464a4a6d45a8bd729e032b4c1b03d9ad69a58c76e0a5c51bd024bfdd68672",
            "signature": "5vwtxvAzTU4q5XVNX6H9JR4bfv+L6Ltbtbmf/8BMHCBCpe/OPfE4rBA5roxUR10SjGHCT/O9Y2PJw0aqbqgVCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "yAC5dhK6YI+EbcYMwzCYGrf5HT3kWlCcbY9GGkWVjM8="
          },
          "build_page": "/inferences/phi-4/mini/bartowski-iq2_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/microsoft_phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W35",
            "2026-W42"
          ],
          "providers": [
            "d68672"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 0.69,
          "context_tokens": 4096,
          "build": "IQ3_XXS (3-bit) by bartowski",
          "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ3_XXS",
          "engine": "ollama-host",
          "cost_usd_per_million": 286.1754,
          "cost_breakdown": {
            "tokens_per_sec": 0.69,
            "usd_per_hour": 0.7097,
            "usd_per_million": 286.1754,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 130.6452
              },
              {
                "resource": "Memory",
                "quantity": "2.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.726,
                "usd_per_million": 2.5055
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 145.1613
              },
              {
                "resource": "Energy",
                "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.417,
                "usd_per_million": 7.8634
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "93XaUH1gQlcOriBiYNWj/r7IvNDXGUwkPw/kJ8bomBa2dFRrSndPa7bisUbIsE3F74Cj5h2oURZmp885nPqFCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "rzUqPv81DXHp4AxZdpcqdvTB21ICTww0FZBTxfrbJAs="
          },
          "build_page": "/inferences/phi-4/mini/bartowski-iq3_xxs-ollama",
          "hub_url": "https://huggingface.co/bartowski/microsoft_phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W42"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": 0.08,
          "context_tokens": 131072,
          "build": "BF16 (16-bit, full precision) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:BF16",
          "engine": "ollama",
          "cost_usd_per_million": 3352.4625,
          "cost_breakdown": {
            "tokens_per_sec": 0.08,
            "usd_per_hour": 0.944,
            "usd_per_million": 3352.4625,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 1917.8116
              },
              {
                "resource": "Memory",
                "quantity": "7.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 5.96,
                "usd_per_million": 76.201
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1278.541
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 79.9088
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "amgW31bT4QjODZ4OEX0ZaMqxLjh7Fd3u67gXsufUi48FYCmK/SpuC7+DJX2ihMbBuPLbtMTtNAijp5ho1ZdPAg==",
            "signing_payload_b64": "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",
            "payload_sha256": "ZX/46eYWosd48NGvUtPQfF5DsyMGVyi5EhzDbLbqe1A="
          },
          "build_page": "/inferences/phi-4/mini/unsloth-bf16-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "BF16 (16-bit, full precision) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:BF16",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-22T09:33:40.320Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/mini/unsloth-bf16-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q5_K_M (5-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q5_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-22T08:56:06.701Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/mini/unsloth-q5_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q8_0 (8-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-22T09:12:21.715Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/mini/unsloth-q8_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-mini-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 4096,
          "build": "BF16 (16-bit, full precision) by microsoft",
          "ref": "microsoft/Phi-4-mini-instruct",
          "engine": "vllm",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-22T08:26:56.172Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/mini/microsoft-bf16-vllm",
          "hub_url": "https://huggingface.co/microsoft/phi-4-mini-instruct",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/microsoft/Phi-4-mini-instruct"
      },
      "builds": 21,
      "variants": [],
      "bottom_line": "205 tokens a second on Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "SmolLM3 3B",
      "made_by": "HuggingFaceTB",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/651e96991b97c9f33d26bde6/e4VK7uW5sTeCYupD0s_ob.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/smollm-3/3b",
      "hub_url": "https://huggingface.co/HuggingFaceTB/SmolLM3-3B",
      "download_gb": 1.8,
      "params_b": 3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 65536,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "reasons about code",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "English",
        "Hindi"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 77.59,
        "context_tokens": 65536,
        "build": "Q4_K_M (4-bit) by bartowski",
        "ref": "hf.co/bartowski/HuggingFaceTB_SmolLM3-3B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.7991,
        "cost_breakdown": {
          "tokens_per_sec": 77.59,
          "usd_per_hour": 0.7818,
          "usd_per_million": 2.7991,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 1.4177
            },
            {
              "resource": "Memory",
              "quantity": "1.8 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.486,
              "usd_per_million": 0.0192
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 1.2889
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0734
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "vNQ+jdq5W0Uyi9s2bF8+JB4DGH/s8siZsoIGHtjGso1zJKQMcC4dyzfKzy36y92nHJaZq+zRdAn9b/y9LHYABg==",
          "signing_payload_b64": "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",
          "payload_sha256": "y9ujbzKC/wGIGaYxwDx9Qi6PfdsibgyamJTLtqree54="
        },
        "build_page": "/inferences/smollm-3/3b/bartowski-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/bartowski/huggingfacetb_smollm3-3b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 77.59,
        "context_tokens": 65536,
        "build": "Q4_K_M (4-bit) by bartowski",
        "ref": "hf.co/bartowski/HuggingFaceTB_SmolLM3-3B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.7991,
        "cost_breakdown": {
          "tokens_per_sec": 77.59,
          "usd_per_hour": 0.7818,
          "usd_per_million": 2.7991,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 1.4177
            },
            {
              "resource": "Memory",
              "quantity": "1.8 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.486,
              "usd_per_million": 0.0192
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 1.2889
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0734
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "vNQ+jdq5W0Uyi9s2bF8+JB4DGH/s8siZsoIGHtjGso1zJKQMcC4dyzfKzy36y92nHJaZq+zRdAn9b/y9LHYABg==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NzcuNTg4NDQ5Njc2MTM4NzIsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0MDI4MTkyOTcwOTcyMjA1LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwNTI0MDUwZTkyOTZkMzEyY2QyODI4YzQyNjE2NWJmZjIzYmFlZWY3NjlhZGE5NzA4MzA5ODgzMmYwNmRhZTY3OSIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbnZpZGlhLzY0ZyIsIm1vZGVsIjoiaGYuY28vYmFydG93c2tpL0h1Z2dpbmdGYWNlVEJfU21vbExNMy0zQi1HR1VGOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo0Nywic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjozNzMuOTA4NzQ2OTk5LCJ0b3RhbF90b2tlbnMiOjI5MDExfX0=",
          "payload_sha256": "y9ujbzKC/wGIGaYxwDx9Qi6PfdsibgyamJTLtqree54="
        },
        "build_page": "/inferences/smollm-3/3b/bartowski-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/bartowski/huggingfacetb_smollm3-3b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 77.59,
          "context_tokens": 65536,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/HuggingFaceTB_SmolLM3-3B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.7991,
          "cost_breakdown": {
            "tokens_per_sec": 77.59,
            "usd_per_hour": 0.7818,
            "usd_per_million": 2.7991,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 1.4177
              },
              {
                "resource": "Memory",
                "quantity": "1.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.486,
                "usd_per_million": 0.0192
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.2889
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0734
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "vNQ+jdq5W0Uyi9s2bF8+JB4DGH/s8siZsoIGHtjGso1zJKQMcC4dyzfKzy36y92nHJaZq+zRdAn9b/y9LHYABg==",
            "signing_payload_b64": "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",
            "payload_sha256": "y9ujbzKC/wGIGaYxwDx9Qi6PfdsibgyamJTLtqree54="
          },
          "build_page": "/inferences/smollm-3/3b/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/huggingfacetb_smollm3-3b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/HuggingFaceTB/SmolLM3-3B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "78 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Llama-3.2-3B-Instruct-uncensored",
      "made_by": "chuanli11",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6223d06e2d6c7fd64778b805/0BgOvZTv-l4UMwjsBxkbE.jpeg",
      "based_on": "Llama 3.2 3B",
      "variant_of": "/inferences/llama-3-2/3b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/llama-3-2/3b~chuanli11-llama-3-2-uncensored",
      "hub_url": "https://huggingface.co/chuanli11/Llama-3.2-3B-Instruct-uncensored",
      "download_gb": 1.5,
      "params_b": 3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 8192,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "returns valid JSON",
        "long answers"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "machine not reported",
        "machine_detail": null,
        "machine_class": "unknown",
        "machines": 1,
        "tokens_per_sec": 46.8,
        "context_tokens": 8192,
        "build": "IQ3_XS (3-bit) by bartowski",
        "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-uncensored-GGUF:IQ3_XS",
        "engine": "ollama-host",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-02T11:31:50.442Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
          "signature": "b/oo2aQS+/Qgttu+TFayPpm6sBm8CTlGlkjB9xs6p/ss82UbiLV9U5sNGGYoufP8zOjfaQtaNqpoMoUc7YLDDQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "Zmy7KqJIynA8GhVSwfxYwPRMiyZwkn1tQ8b1VVzb4g8="
        },
        "build_page": "/inferences/llama-3-2/3b~chuanli11-llama-3-2-uncensored/bartowski-iq3_xs-ollama",
        "hub_url": "https://huggingface.co/bartowski/llama-3.2-3b-instruct-uncensored-gguf",
        "weeks": [
          "2026-W35"
        ],
        "providers": [
          "afbb8a"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W35"
      ],
      "measured_on": [
        {
          "machine": "machine not reported",
          "machine_detail": null,
          "machine_class": "unknown",
          "machines": 1,
          "tokens_per_sec": 46.8,
          "context_tokens": 8192,
          "build": "IQ3_XS (3-bit) by bartowski",
          "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-uncensored-GGUF:IQ3_XS",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "b/oo2aQS+/Qgttu+TFayPpm6sBm8CTlGlkjB9xs6p/ss82UbiLV9U5sNGGYoufP8zOjfaQtaNqpoMoUc7YLDDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "Zmy7KqJIynA8GhVSwfxYwPRMiyZwkn1tQ8b1VVzb4g8="
          },
          "build_page": "/inferences/llama-3-2/3b~chuanli11-llama-3-2-uncensored/bartowski-iq3_xs-ollama",
          "hub_url": "https://huggingface.co/bartowski/llama-3.2-3b-instruct-uncensored-gguf",
          "weeks": [
            "2026-W35"
          ],
          "providers": [
            "afbb8a"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": "https://huggingface.co/chuanli11/Llama-3.2-3B-Instruct-uncensored"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "47 tokens a second on a machine not reported. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "MiniCPM4.1 8B",
      "made_by": "openbmb",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1670387859384-633fe7784b362488336bbfad.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/minicpm-4-1/8b",
      "hub_url": "https://huggingface.co/openbmb/MiniCPM4.1-8B-GGUF",
      "download_gb": 4.6,
      "params_b": 8,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 65536,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "reasons about code",
        "shows its reasoning",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "English",
        "Hindi",
        "Malayalam",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 45.81,
        "context_tokens": 65536,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM4.1-8B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 4.7923,
        "cost_breakdown": {
          "tokens_per_sec": 45.81,
          "usd_per_hour": 0.7904,
          "usd_per_million": 4.7923,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 2.4011
            },
            {
              "resource": "Memory",
              "quantity": "4.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 3.853,
              "usd_per_million": 0.0841
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 2.1828
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.1243
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "oKC0Iux/EPIFdOmaybf/XLlvTxWIFqlW5yz2G9gKZu14QMOxJBpUFBk3a/k7h9Rkl1sqdLvs3x06RO6AriNWBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "OKHWYI52qLuV/meKsWPrlTFLbpDvo/lrETS1IAnIJYE="
        },
        "build_page": "/inferences/minicpm-4-1/8b/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm4.1-8b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 45.81,
        "context_tokens": 65536,
        "build": "Q4_K_M (4-bit) by openbmb",
        "ref": "hf.co/openbmb/MiniCPM4.1-8B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 4.7923,
        "cost_breakdown": {
          "tokens_per_sec": 45.81,
          "usd_per_hour": 0.7904,
          "usd_per_million": 4.7923,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 2.4011
            },
            {
              "resource": "Memory",
              "quantity": "4.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 3.853,
              "usd_per_million": 0.0841
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 2.1828
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.1243
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "oKC0Iux/EPIFdOmaybf/XLlvTxWIFqlW5yz2G9gKZu14QMOxJBpUFBk3a/k7h9Rkl1sqdLvs3x06RO6AriNWBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "OKHWYI52qLuV/meKsWPrlTFLbpDvo/lrETS1IAnIJYE="
        },
        "build_page": "/inferences/minicpm-4-1/8b/openbmb-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/openbmb/minicpm4.1-8b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 45.81,
          "context_tokens": 65536,
          "build": "Q4_K_M (4-bit) by openbmb",
          "ref": "hf.co/openbmb/MiniCPM4.1-8B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 4.7923,
          "cost_breakdown": {
            "tokens_per_sec": 45.81,
            "usd_per_hour": 0.7904,
            "usd_per_million": 4.7923,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 2.4011
              },
              {
                "resource": "Memory",
                "quantity": "4.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.853,
                "usd_per_million": 0.0841
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.1828
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.1243
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "oKC0Iux/EPIFdOmaybf/XLlvTxWIFqlW5yz2G9gKZu14QMOxJBpUFBk3a/k7h9Rkl1sqdLvs3x06RO6AriNWBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "OKHWYI52qLuV/meKsWPrlTFLbpDvo/lrETS1IAnIJYE="
          },
          "build_page": "/inferences/minicpm-4-1/8b/openbmb-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/openbmb/minicpm4.1-8b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openbmb/MiniCPM4.1-8B-GGUF"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "46 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Sarvam-M",
      "made_by": "sarvamai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/60270a7c32856987162c641a/umd13GCWVijwTDGZzw3q-.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/sarvam/m",
      "hub_url": "https://huggingface.co/sarvamai/sarvam-m",
      "download_gb": 13.3,
      "params_b": 24,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 8192,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "reasons about code",
        "shows its reasoning",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": 15.61,
        "context_tokens": 8192,
        "build": "Q4_K_M (4-bit) by bartowski",
        "ref": "hf.co/bartowski/sarvamai_sarvam-m-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 9.3329,
        "cost_breakdown": {
          "tokens_per_sec": 15.61,
          "usd_per_hour": 0.5245,
          "usd_per_million": 9.3329,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 1.9216
            },
            {
              "resource": "Memory",
              "quantity": "13.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 11.124,
              "usd_per_million": 0.7125
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 6.4052
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.2936
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "IiaX6zXGosC9aNd7a4vC5mBOFHuQ4TtpW1L7qTu4LCS4OYJpvvFmgn+fM6O1zpm3+4/gokwF2cXFjaR3htSKCQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "FdGHr57sQjHOTEt8WGZd23fANyfBF53d8ZZhNz8q0BY="
        },
        "build_page": "/inferences/sarvam/m/bartowski-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/bartowski/sarvamai_sarvam-m-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": 15.61,
        "context_tokens": 8192,
        "build": "Q4_K_M (4-bit) by bartowski",
        "ref": "hf.co/bartowski/sarvamai_sarvam-m-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 9.3329,
        "cost_breakdown": {
          "tokens_per_sec": 15.61,
          "usd_per_hour": 0.5245,
          "usd_per_million": 9.3329,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 1.9216
            },
            {
              "resource": "Memory",
              "quantity": "13.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 11.124,
              "usd_per_million": 0.7125
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 6.4052
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.2936
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "IiaX6zXGosC9aNd7a4vC5mBOFHuQ4TtpW1L7qTu4LCS4OYJpvvFmgn+fM6O1zpm3+4/gokwF2cXFjaR3htSKCQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "FdGHr57sQjHOTEt8WGZd23fANyfBF53d8ZZhNz8q0BY="
        },
        "build_page": "/inferences/sarvam/m/bartowski-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/bartowski/sarvamai_sarvam-m-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 15.61,
          "context_tokens": 8192,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/sarvamai_sarvam-m-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 9.3329,
          "cost_breakdown": {
            "tokens_per_sec": 15.61,
            "usd_per_hour": 0.5245,
            "usd_per_million": 9.3329,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.9216
              },
              {
                "resource": "Memory",
                "quantity": "13.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 11.124,
                "usd_per_million": 0.7125
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 6.4052
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.2936
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "IiaX6zXGosC9aNd7a4vC5mBOFHuQ4TtpW1L7qTu4LCS4OYJpvvFmgn+fM6O1zpm3+4/gokwF2cXFjaR3htSKCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "FdGHr57sQjHOTEt8WGZd23fANyfBF53d8ZZhNz8q0BY="
          },
          "build_page": "/inferences/sarvam/m/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/sarvamai_sarvam-m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/sarvamai/sarvam-m"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "16 tokens a second on Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Krutrim-2 Instruct",
      "made_by": "krutrim-ai-labs",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/679e056019b9b76439a55a33/xJACZDT7g3veiRGBvfXOK.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/krutrim/2",
      "hub_url": "https://huggingface.co/krutrim-ai-labs/Krutrim-2-instruct",
      "download_gb": 7,
      "params_b": 12,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 65536,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu",
        "Hindi"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 14.49,
        "context_tokens": 32768,
        "build": "Q4_K_M (4-bit) by bartowski",
        "ref": "hf.co/bartowski/krutrim-ai-labs_Krutrim-2-instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 15.2823,
        "cost_breakdown": {
          "tokens_per_sec": 14.49,
          "usd_per_hour": 0.7974,
          "usd_per_million": 15.2823,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 7.5895
            },
            {
              "resource": "Memory",
              "quantity": "7.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 5.802,
              "usd_per_million": 0.4003
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 6.8996
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.3929
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "z44e0c3eU9BSOGCqW9ua6PRQDVs1iqeYn9wq7ahuJ9mQfIlu92QHj59qz16RACFmfZ7K/Lm4PdHce4DJzwEFBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "tvRXXqsyBiV/C6V0yFMqPnUmgvRxaGwp8JqF6Ipz86w="
        },
        "build_page": "/inferences/krutrim/2/bartowski-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/bartowski/krutrim-ai-labs_krutrim-2-instruct-gguf",
        "weeks": [
          "2026-W43",
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": 13.56,
        "context_tokens": 65536,
        "build": "Q4_K_M (4-bit) by bartowski",
        "ref": "hf.co/bartowski/krutrim-ai-labs_Krutrim-2-instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 10.3505,
        "cost_breakdown": {
          "tokens_per_sec": 13.56,
          "usd_per_hour": 0.5054,
          "usd_per_million": 10.3505,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 2.2119
            },
            {
              "resource": "Memory",
              "quantity": "7.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 5.802,
              "usd_per_million": 0.4278
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 7.3729
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.3379
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "Ck44BL42uOhTnx5nwRTcDGqfSVXDnkqa5Et+jgp7GjEgU7G36ACxA+9AYdZVDfrIM/T2TTGrHYD+Y1XM4sDeDg==",
          "signing_payload_b64": "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",
          "payload_sha256": "h5TODmxeM82uG4raHdhz2qwPrt/C3iegbEyTiCAf550="
        },
        "build_page": "/inferences/krutrim/2/bartowski-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/bartowski/krutrim-ai-labs_krutrim-2-instruct-gguf",
        "weeks": [
          "2026-W43",
          "2026-W45"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "machine_kinds": 2,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W43",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 14.49,
          "context_tokens": 32768,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/krutrim-ai-labs_Krutrim-2-instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 15.2823,
          "cost_breakdown": {
            "tokens_per_sec": 14.49,
            "usd_per_hour": 0.7974,
            "usd_per_million": 15.2823,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 7.5895
              },
              {
                "resource": "Memory",
                "quantity": "7.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 5.802,
                "usd_per_million": 0.4003
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 6.8996
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.3929
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "z44e0c3eU9BSOGCqW9ua6PRQDVs1iqeYn9wq7ahuJ9mQfIlu92QHj59qz16RACFmfZ7K/Lm4PdHce4DJzwEFBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "tvRXXqsyBiV/C6V0yFMqPnUmgvRxaGwp8JqF6Ipz86w="
          },
          "build_page": "/inferences/krutrim/2/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/krutrim-ai-labs_krutrim-2-instruct-gguf",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 13.56,
          "context_tokens": 65536,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/krutrim-ai-labs_Krutrim-2-instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 10.3505,
          "cost_breakdown": {
            "tokens_per_sec": 13.56,
            "usd_per_hour": 0.5054,
            "usd_per_million": 10.3505,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.2119
              },
              {
                "resource": "Memory",
                "quantity": "7.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 5.802,
                "usd_per_million": 0.4278
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 7.3729
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.3379
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "Ck44BL42uOhTnx5nwRTcDGqfSVXDnkqa5Et+jgp7GjEgU7G36ACxA+9AYdZVDfrIM/T2TTGrHYD+Y1XM4sDeDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "h5TODmxeM82uG4raHdhz2qwPrt/C3iegbEyTiCAf550="
          },
          "build_page": "/inferences/krutrim/2/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/krutrim-ai-labs_krutrim-2-instruct-gguf",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Krutrim Community License Agreement Version 1.0 licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/krutrim-ai-labs/Krutrim-2-instruct"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "14 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Granite 4.0 1B",
      "made_by": "ibm-granite",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/639bcaa2445b133a4e942436/CEW-OjXkRkDNmTxSu8Egh.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/granite-4-0/1b",
      "hub_url": "https://huggingface.co/ibm-granite/granite-4.0-1b",
      "download_gb": 1,
      "params_b": 1,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "writes code",
        "returns valid JSON",
        "long answers"
      ],
      "languages_passed": [
        "English",
        "Malayalam",
        "Tamil",
        "Telugu",
        "Dutch",
        "Russian"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 2,
        "tokens_per_sec": 11.37,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by ibm-granite",
        "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.744,
        "cost_breakdown": {
          "tokens_per_sec": 11.37,
          "usd_per_hour": 0.1124,
          "usd_per_million": 2.744,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 2.6376
            },
            {
              "resource": "Memory",
              "quantity": "1.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.794,
              "usd_per_million": 0.0698
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0367
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "nqLXOqxDriaxROVdKEjJzh4w1AwdH4z9t9XISaALgYLFWL/lVJv/M/CrIMEy3TEDbtrarVVB0DKziBwXiteQDQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "z76Hz4HoYUep4hYU8qw84UMdoj1i6aybsAqYhxCvglw="
        },
        "build_page": "/inferences/granite-4-0/1b/ibm-granite-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/ibm-granite/granite-4.0-1b-gguf",
        "weeks": [
          "2026-W43",
          "2026-W44"
        ],
        "providers": [
          "e61d3b",
          "acbe75"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 2,
        "tokens_per_sec": 11.37,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by ibm-granite",
        "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.744,
        "cost_breakdown": {
          "tokens_per_sec": 11.37,
          "usd_per_hour": 0.1124,
          "usd_per_million": 2.744,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 2.6376
            },
            {
              "resource": "Memory",
              "quantity": "1.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.794,
              "usd_per_million": 0.0698
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0367
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "nqLXOqxDriaxROVdKEjJzh4w1AwdH4z9t9XISaALgYLFWL/lVJv/M/CrIMEy3TEDbtrarVVB0DKziBwXiteQDQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "z76Hz4HoYUep4hYU8qw84UMdoj1i6aybsAqYhxCvglw="
        },
        "build_page": "/inferences/granite-4-0/1b/ibm-granite-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/ibm-granite/granite-4.0-1b-gguf",
        "weeks": [
          "2026-W43",
          "2026-W44"
        ],
        "providers": [
          "e61d3b",
          "acbe75"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 7,
      "weeks_measured": [
        "2026-W43",
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 2,
          "tokens_per_sec": 11.37,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by ibm-granite",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.744,
          "cost_breakdown": {
            "tokens_per_sec": 11.37,
            "usd_per_hour": 0.1124,
            "usd_per_million": 2.744,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.6376
              },
              {
                "resource": "Memory",
                "quantity": "1.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.794,
                "usd_per_million": 0.0698
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0367
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "nqLXOqxDriaxROVdKEjJzh4w1AwdH4z9t9XISaALgYLFWL/lVJv/M/CrIMEy3TEDbtrarVVB0DKziBwXiteQDQ==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTEuMzc0MTQ5NzAwMDk3MDI1LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5MzQwNTkxNzI5ODY0ODg5MSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMGVlMzVkYzQ2YTIyMTJjYjMxNDQ0Y2E0YzE5NjY5YWM4ZDk3YjA5MjUyMGI4YzllNGQyOWFiNDhkOWVlNjFkM2IiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L25vbmUvMzJnIiwibW9kZWwiOiJoZi5jby9pYm0tZ3Jhbml0ZS9ncmFuaXRlLTQuMC0xYi1HR1VGOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo1NSwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjoyOTcuODY4NDE5OTk4OTk5OTcsInRvdGFsX3Rva2VucyI6MzM4OH19",
            "payload_sha256": "z76Hz4HoYUep4hYU8qw84UMdoj1i6aybsAqYhxCvglw="
          },
          "build_page": "/inferences/granite-4-0/1b/ibm-granite-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/ibm-granite/granite-4.0-1b-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44"
          ],
          "providers": [
            "e61d3b",
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 8.75,
          "context_tokens": 4096,
          "build": "Q4_0 (4-bit) by ibm-granite",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 3.5635,
          "cost_breakdown": {
            "tokens_per_sec": 8.75,
            "usd_per_hour": 0.1122,
            "usd_per_million": 3.5635,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 3.4295
              },
              {
                "resource": "Memory",
                "quantity": "0.9 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.756,
                "usd_per_million": 0.0864
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0477
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "GEOfxTYMpxeor6q4+IwRuKk/oPADCFK8sbl3+h3qNUu3ESeGs1KWE6HdgleZZr49dlBFsyc5bqC5VH3Ed6PsBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "5pfYcU6+ABp+AEDd0l4HeX6Lhhj9cVi40eshalQ7fc4="
          },
          "build_page": "/inferences/granite-4-0/1b/ibm-granite-q4_0-ollama",
          "hub_url": "https://huggingface.co/ibm-granite/granite-4.0-1b-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 7.98,
          "context_tokens": 4096,
          "build": "Q3_K_S (3-bit) by ibm-granite",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 3.8867,
          "cost_breakdown": {
            "tokens_per_sec": 7.98,
            "usd_per_hour": 0.1117,
            "usd_per_million": 3.8867,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 3.7581
              },
              {
                "resource": "Memory",
                "quantity": "0.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.61,
                "usd_per_million": 0.0764
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0522
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "gxcNkCZgHM72QnBInh+ro9oljDIeN9A+P2BwWMmbEHZcOwWxMbT/tkjZ1qR3zGnuanGOdXYhR8+n/bLShrA+Bw==",
            "signing_payload_b64": "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",
            "payload_sha256": "eo6WyNhxO2RPRQZj5nnbQpqREXZh+HQ4ucVSgkICrlk="
          },
          "build_page": "/inferences/granite-4-0/1b/ibm-granite-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/ibm-granite/granite-4.0-1b-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 7.82,
          "context_tokens": null,
          "build": "Q2_K (2-bit) by ibm-granite",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q2_K",
          "engine": "ollama-host",
          "cost_usd_per_million": 3.9609,
          "cost_breakdown": {
            "tokens_per_sec": 7.82,
            "usd_per_hour": 0.1115,
            "usd_per_million": 3.9609,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 3.838
              },
              {
                "resource": "Memory",
                "quantity": "0.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.544,
                "usd_per_million": 0.0696
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0533
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "pI2VwoF8V8R/ThMzzxlzdzd2ORlDfk9hMwb3Sj3B8VmHUbyhuwf1Ar9jxmYGVEtZXHxD+vM3PBtdrTOuY3DXCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "qXp3RDYKZBeCgKtHUmVkt9wJbxxTPCENWgTFNWmXqVc="
          },
          "build_page": "/inferences/granite-4-0/1b/ibm-granite-q2_k-ollama",
          "hub_url": "https://huggingface.co/ibm-granite/granite-4.0-1b-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 7.23,
          "context_tokens": 4096,
          "build": "Q3_K_L (3-bit) by ibm-granite",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "cost_usd_per_million": 4.3044,
          "cost_breakdown": {
            "tokens_per_sec": 7.23,
            "usd_per_hour": 0.1121,
            "usd_per_million": 4.3044,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 4.1473
              },
              {
                "resource": "Memory",
                "quantity": "0.9 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.719,
                "usd_per_million": 0.0994
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0576
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "tqOKISwjQs0EocUz7i5Z0jPjTAwZHUxgsxZkuVU3fsOH0uvqyE3aD9LbcDAvCfr6scIn/pPRO32ALEQZrRL7DA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6Ny4yMzM2MDc0NDk1MDE0NzEsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzNDA1OTE3Mjk4NjQ4ODkxLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwZWUzNWRjNDZhMjIxMmNiMzE0NDRjYTRjMTk2NjlhYzhkOTdiMDkyNTIwYjhjOWU0ZDI5YWI0OGQ5ZWU2MWQzYiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbm9uZS8zMmciLCJtb2RlbCI6ImhmLmNvL2libS1ncmFuaXRlL2dyYW5pdGUtNC4wLTFiLUdHVUY6UTNfS19MIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjU3LCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjU2Ny40OTAwMTQ5OTY5OTk5LCJ0b3RhbF90b2tlbnMiOjQxMDV9fQ==",
            "payload_sha256": "8DTR04zMbtp0f+Op6R5gLMeyTnnG80GEGvOKyeeR5P0="
          },
          "build_page": "/inferences/granite-4-0/1b/ibm-granite-q3_k_l-ollama",
          "hub_url": "https://huggingface.co/ibm-granite/granite-4.0-1b-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 4.71,
          "context_tokens": 4096,
          "build": "BF16 (16-bit, full precision) by ibm-granite",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:BF16",
          "engine": "ollama-host",
          "cost_usd_per_million": 6.9928,
          "cost_breakdown": {
            "tokens_per_sec": 4.71,
            "usd_per_hour": 0.1186,
            "usd_per_million": 6.9928,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.3661
              },
              {
                "resource": "Memory",
                "quantity": "3.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 2.536,
                "usd_per_million": 0.5382
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0885
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "Q5MaNrs4bUmuY7+o48kTpDfNpEdtCywAvzEVRUl9Jadv2QI7VGC63p91qww8F5G48/MLZp64jov+tWCq82G4Dg==",
            "signing_payload_b64": "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",
            "payload_sha256": "CHwLS3E2726+k6/CPWzr4fHNlFBUt5+NBi/nBWqpHkM="
          },
          "build_page": "/inferences/granite-4-0/1b/ibm-granite-bf16-ollama",
          "hub_url": "https://huggingface.co/ibm-granite/granite-4.0-1b-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/ibm-granite/granite-4.0-1b"
      },
      "builds": 6,
      "variants": [],
      "bottom_line": "11 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Qwen 3 VL-2B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3/vl-2b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct",
      "download_gb": 1,
      "params_b": 2,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "calls tools",
        "returns valid JSON",
        "reads images",
        "counts objects"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 9.24,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by Qwen",
        "ref": "hf.co/Qwen/Qwen3-VL-2B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 3.3842,
        "cost_breakdown": {
          "tokens_per_sec": 9.24,
          "usd_per_hour": 0.1126,
          "usd_per_million": 3.3842,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 3.2461
            },
            {
              "resource": "Memory",
              "quantity": "1.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.859,
              "usd_per_million": 0.0929
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0451
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD EPYC 7B12",
            "model_raw": "AMD EPYC 7B12",
            "clock_ghz": null,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 512,
            "l3_mb": 16,
            "single_core_hashes_per_sec": 1460.2,
            "multi_core_hashes_per_sec": 3246.1,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T01:38:20.212Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 6.3,
            "latency_ns": 21.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T01:38:21.135Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4982.5,
            "write_mb_s": 179.3,
            "random_read_iops": 476287,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-15T01:38:21.741Z",
            "measured_in": "TW_110",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-15T06:44:44.996Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
          "signature": "SgSuH+BqllkOZEseVft005CqxKyZ6F6Kd1oyNJr+6aSjhisCsw7f8O3bDilr8quGdp+No2dAPXiCKJ6e+em2Bw==",
          "signing_payload_b64": "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",
          "payload_sha256": "nrdAB6QxpwN/9Cla2yZ7FqVDJb4NpYQ6snBietKbRY4="
        },
        "build_page": "/inferences/qwen-3/vl-2b/qwen-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/qwen/qwen3-vl-2b-instruct-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "662a4e"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 9.24,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by Qwen",
        "ref": "hf.co/Qwen/Qwen3-VL-2B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 3.3842,
        "cost_breakdown": {
          "tokens_per_sec": 9.24,
          "usd_per_hour": 0.1126,
          "usd_per_million": 3.3842,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 3.2461
            },
            {
              "resource": "Memory",
              "quantity": "1.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.859,
              "usd_per_million": 0.0929
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0451
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD EPYC 7B12",
            "model_raw": "AMD EPYC 7B12",
            "clock_ghz": null,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 512,
            "l3_mb": 16,
            "single_core_hashes_per_sec": 1460.2,
            "multi_core_hashes_per_sec": 3246.1,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T01:38:20.212Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 6.3,
            "latency_ns": 21.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T01:38:21.135Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4982.5,
            "write_mb_s": 179.3,
            "random_read_iops": 476287,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-15T01:38:21.741Z",
            "measured_in": "TW_110",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-15T06:44:44.996Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
          "signature": "SgSuH+BqllkOZEseVft005CqxKyZ6F6Kd1oyNJr+6aSjhisCsw7f8O3bDilr8quGdp+No2dAPXiCKJ6e+em2Bw==",
          "signing_payload_b64": "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",
          "payload_sha256": "nrdAB6QxpwN/9Cla2yZ7FqVDJb4NpYQ6snBietKbRY4="
        },
        "build_page": "/inferences/qwen-3/vl-2b/qwen-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/qwen/qwen3-vl-2b-instruct-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "662a4e"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 9.24,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by Qwen",
          "ref": "hf.co/Qwen/Qwen3-VL-2B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 3.3842,
          "cost_breakdown": {
            "tokens_per_sec": 9.24,
            "usd_per_hour": 0.1126,
            "usd_per_million": 3.3842,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 3.2461
              },
              {
                "resource": "Memory",
                "quantity": "1.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.859,
                "usd_per_million": 0.0929
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0451
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:44:44.996Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "SgSuH+BqllkOZEseVft005CqxKyZ6F6Kd1oyNJr+6aSjhisCsw7f8O3bDilr8quGdp+No2dAPXiCKJ6e+em2Bw==",
            "signing_payload_b64": "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",
            "payload_sha256": "nrdAB6QxpwN/9Cla2yZ7FqVDJb4NpYQ6snBietKbRY4="
          },
          "build_page": "/inferences/qwen-3/vl-2b/qwen-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/qwen/qwen3-vl-2b-instruct-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "9.2 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Kimi-Linear-Reap-35B-A3B-Instruct",
      "made_by": "cerebras",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1678752782877-640fbb91bd54b1efde3678b2.png",
      "based_on": "Kimi Linear",
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/kimi-linear/35b-a3b~cerebras-reap",
      "hub_url": "https://huggingface.co/cerebras/Kimi-Linear-REAP-35B-A3B-Instruct",
      "download_gb": 19.8,
      "params_b": 35,
      "active_params_b": 3,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Kannada",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 4.75,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by mradermacher",
        "ref": "hf.co/mradermacher/Kimi-Linear-REAP-35B-A3B-Instruct-i1-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 9.8821,
        "cost_breakdown": {
          "tokens_per_sec": 4.75,
          "usd_per_hour": 0.1689,
          "usd_per_million": 9.8821,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 6.3187
            },
            {
              "resource": "Memory",
              "quantity": "19.8 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 16.501,
              "usd_per_million": 3.4755
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0878
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "O50OXK3rPvrKCrLG8lS0x49NkRYMG3tasYGmx4OSND4NUQjsWXZzfn83hAm55qHamCDhPTZdsE11kS63DP+yAQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "+cfCglecb3AakW6wGod4BVL5rY0/BWZJFkkBhNI+uck="
        },
        "build_page": "/inferences/kimi-linear/35b-a3b~cerebras-reap/mradermacher-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/mradermacher/kimi-linear-reap-35b-a3b-instruct-i1-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 4.75,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by mradermacher",
        "ref": "hf.co/mradermacher/Kimi-Linear-REAP-35B-A3B-Instruct-i1-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 9.8821,
        "cost_breakdown": {
          "tokens_per_sec": 4.75,
          "usd_per_hour": 0.1689,
          "usd_per_million": 9.8821,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 6.3187
            },
            {
              "resource": "Memory",
              "quantity": "19.8 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 16.501,
              "usd_per_million": 3.4755
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0878
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "O50OXK3rPvrKCrLG8lS0x49NkRYMG3tasYGmx4OSND4NUQjsWXZzfn83hAm55qHamCDhPTZdsE11kS63DP+yAQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "+cfCglecb3AakW6wGod4BVL5rY0/BWZJFkkBhNI+uck="
        },
        "build_page": "/inferences/kimi-linear/35b-a3b~cerebras-reap/mradermacher-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/mradermacher/kimi-linear-reap-35b-a3b-instruct-i1-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 4.75,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by mradermacher",
          "ref": "hf.co/mradermacher/Kimi-Linear-REAP-35B-A3B-Instruct-i1-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 9.8821,
          "cost_breakdown": {
            "tokens_per_sec": 4.75,
            "usd_per_hour": 0.1689,
            "usd_per_million": 9.8821,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.3187
              },
              {
                "resource": "Memory",
                "quantity": "19.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 16.501,
                "usd_per_million": 3.4755
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0878
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "O50OXK3rPvrKCrLG8lS0x49NkRYMG3tasYGmx4OSND4NUQjsWXZzfn83hAm55qHamCDhPTZdsE11kS63DP+yAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "+cfCglecb3AakW6wGod4BVL5rY0/BWZJFkkBhNI+uck="
          },
          "build_page": "/inferences/kimi-linear/35b-a3b~cerebras-reap/mradermacher-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/mradermacher/kimi-linear-reap-35b-a3b-instruct-i1-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/cerebras/Kimi-Linear-REAP-35B-A3B-Instruct"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "4.8 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Phi 4 Reasoning-Plus",
      "made_by": "microsoft",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1583646260758-5e64858c87403103f9f1055d.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/phi-4/reasoning-plus",
      "hub_url": "https://huggingface.co/microsoft/Phi-4-reasoning-plus",
      "download_gb": 8.4,
      "params_b": 15,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": 32768,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "reasons about code",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "English",
        "Hindi",
        "Kannada",
        "Bengali",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "os": "Mac",
          "os_version": "macOS 26.5.2",
          "cpus": 10,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "Apple",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": null,
          "host": null,
          "vendor_org": "apple",
          "host_org": null,
          "key": "darwin|arm64|10|16|apple|0|||"
        },
        "machine_class": "darwin|arm64|10|16|apple|0|||",
        "machines": 1,
        "tokens_per_sec": 3.26,
        "context_tokens": null,
        "build": "Q6_K (6-bit) by lmstudio-community",
        "ref": "hf.co/lmstudio-community/Phi-4-reasoning-plus-GGUF:Q6_K",
        "engine": "ollama-host",
        "cost_usd_per_million": 60.0111,
        "cost_breakdown": {
          "tokens_per_sec": 3.26,
          "usd_per_hour": 0.7035,
          "usd_per_million": 60.0111,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "9 of 10 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 90,
              "usd_per_million": 27.6383
            },
            {
              "resource": "Memory",
              "quantity": "0.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0,
              "usd_per_million": 0
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 30.7093
            },
            {
              "resource": "Energy",
              "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.417,
              "usd_per_million": 1.6635
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Apple M4",
            "model_raw": "Apple M4",
            "clock_ghz": null,
            "cores_logical": 10,
            "cores_physical": 10,
            "threads_per_core": 1,
            "sockets": 1,
            "l2_kb": 4096,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2850.2,
            "multi_core_hashes_per_sec": 22838.7,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-02T08:29:50.654Z",
            "measured_in": "IN_110054",
            "signed": true
          },
          "memory": {
            "total_gb": 16,
            "bandwidth_gb_s": 42.7,
            "latency_ns": 6.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-02T08:29:51.180Z",
            "measured_in": "IN_110054",
            "signed": true
          },
          "storage": {
            "read_mb_s": 13543.1,
            "write_mb_s": 1696,
            "random_read_iops": 1733096,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-02T08:29:51.427Z",
            "measured_in": "IN_110054",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-02T11:31:50.442Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
          "signature": "hFDo9wK6MhBqmtCWuWYsxVFRTfQb0wKoKFQC4brRr2vDevawclDeswNcORKGualHahw4WS7s09YQnv/OoCCQAQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "1WnnjF0uxnidjiRXm9j8zaDZQjU/lElGizDI0ATDZ/Q="
        },
        "build_page": "/inferences/phi-4/reasoning-plus/lmstudio-community-q6_k-ollama",
        "hub_url": "https://huggingface.co/lmstudio-community/phi-4-reasoning-plus-gguf",
        "weeks": [
          "2026-W38"
        ],
        "providers": [
          "afbb8a"
        ]
      },
      "cheapest": {
        "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "os": "Mac",
          "os_version": "macOS 26.5.2",
          "cpus": 10,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "Apple",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": null,
          "host": null,
          "vendor_org": "apple",
          "host_org": null,
          "key": "darwin|arm64|10|16|apple|0|||"
        },
        "machine_class": "darwin|arm64|10|16|apple|0|||",
        "machines": 1,
        "tokens_per_sec": 3.26,
        "context_tokens": null,
        "build": "Q6_K (6-bit) by lmstudio-community",
        "ref": "hf.co/lmstudio-community/Phi-4-reasoning-plus-GGUF:Q6_K",
        "engine": "ollama-host",
        "cost_usd_per_million": 60.0111,
        "cost_breakdown": {
          "tokens_per_sec": 3.26,
          "usd_per_hour": 0.7035,
          "usd_per_million": 60.0111,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "9 of 10 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 90,
              "usd_per_million": 27.6383
            },
            {
              "resource": "Memory",
              "quantity": "0.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0,
              "usd_per_million": 0
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 30.7093
            },
            {
              "resource": "Energy",
              "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.417,
              "usd_per_million": 1.6635
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Apple M4",
            "model_raw": "Apple M4",
            "clock_ghz": null,
            "cores_logical": 10,
            "cores_physical": 10,
            "threads_per_core": 1,
            "sockets": 1,
            "l2_kb": 4096,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2850.2,
            "multi_core_hashes_per_sec": 22838.7,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-02T08:29:50.654Z",
            "measured_in": "IN_110054",
            "signed": true
          },
          "memory": {
            "total_gb": 16,
            "bandwidth_gb_s": 42.7,
            "latency_ns": 6.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-02T08:29:51.180Z",
            "measured_in": "IN_110054",
            "signed": true
          },
          "storage": {
            "read_mb_s": 13543.1,
            "write_mb_s": 1696,
            "random_read_iops": 1733096,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-02T08:29:51.427Z",
            "measured_in": "IN_110054",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-02T11:31:50.442Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
          "signature": "hFDo9wK6MhBqmtCWuWYsxVFRTfQb0wKoKFQC4brRr2vDevawclDeswNcORKGualHahw4WS7s09YQnv/OoCCQAQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "1WnnjF0uxnidjiRXm9j8zaDZQjU/lElGizDI0ATDZ/Q="
        },
        "build_page": "/inferences/phi-4/reasoning-plus/lmstudio-community-q6_k-ollama",
        "hub_url": "https://huggingface.co/lmstudio-community/phi-4-reasoning-plus-gguf",
        "weeks": [
          "2026-W38"
        ],
        "providers": [
          "afbb8a"
        ]
      },
      "machine_kinds": 2,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W38",
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 3.26,
          "context_tokens": null,
          "build": "Q6_K (6-bit) by lmstudio-community",
          "ref": "hf.co/lmstudio-community/Phi-4-reasoning-plus-GGUF:Q6_K",
          "engine": "ollama-host",
          "cost_usd_per_million": 60.0111,
          "cost_breakdown": {
            "tokens_per_sec": 3.26,
            "usd_per_hour": 0.7035,
            "usd_per_million": 60.0111,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 27.6383
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 30.7093
              },
              {
                "resource": "Energy",
                "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.417,
                "usd_per_million": 1.6635
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "hFDo9wK6MhBqmtCWuWYsxVFRTfQb0wKoKFQC4brRr2vDevawclDeswNcORKGualHahw4WS7s09YQnv/OoCCQAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "1WnnjF0uxnidjiRXm9j8zaDZQjU/lElGizDI0ATDZ/Q="
          },
          "build_page": "/inferences/phi-4/reasoning-plus/lmstudio-community-q6_k-ollama",
          "hub_url": "https://huggingface.co/lmstudio-community/phi-4-reasoning-plus-gguf",
          "weeks": [
            "2026-W38"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by lmstudio-community",
          "ref": "hf.co/lmstudio-community/Phi-4-reasoning-plus-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-15T04:44:02.843Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/reasoning-plus/lmstudio-community-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/lmstudio-community/phi-4-reasoning-plus-gguf",
          "weeks": [
            "2026-W38"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 32768,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Phi-4-reasoning-plus-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T18:52:51.415Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/phi-4/reasoning-plus/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/phi-4-reasoning-plus-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/microsoft/Phi-4-reasoning-plus"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "3.3 tokens a second on Mac · Apple silicon · 10 CPUs · 16 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill",
      "made_by": "TeichAI",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6837935ac3b7ffe0d2559ce9/-AxyvV4wfUY8uo87kNKkK.png",
      "based_on": "DeepSeek V3.2",
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/deepseek-v3-2/devstral-small-2505-speciale-distill~teichai",
      "hub_url": "https://huggingface.co/TeichAI/Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill",
      "download_gb": 9.2,
      "params_b": 21,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Kannada",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 0.36,
        "context_tokens": 4096,
        "build": "IQ3_XS (3-bit) by TeichAI",
        "ref": "hf.co/TeichAI/Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill-GGUF:IQ3_XS",
        "engine": "ollama-host",
        "cost_usd_per_million": 105.2094,
        "cost_breakdown": {
          "tokens_per_sec": 0.36,
          "usd_per_hour": 0.1372,
          "usd_per_million": 105.2094,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 82.829
            },
            {
              "resource": "Memory",
              "quantity": "9.2 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 7.689,
              "usd_per_million": 21.2291
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 1.1513
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD EPYC 7B12",
            "model_raw": "AMD EPYC 7B12",
            "clock_ghz": null,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 512,
            "l3_mb": 16,
            "single_core_hashes_per_sec": 1460.2,
            "multi_core_hashes_per_sec": 3246.1,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T01:38:20.212Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 6.3,
            "latency_ns": 21.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T01:38:21.135Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4982.5,
            "write_mb_s": 179.3,
            "random_read_iops": 476287,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-15T01:38:21.741Z",
            "measured_in": "TW_110",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-15T06:44:44.996Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
          "signature": "YKSegtdbk3BFhfs4/FH9yVX6NFoPLjmSlaFS0ZpkFUuDBmD4srmGn7dgiwir/KQtarmq6qTVAiZWCTonIjYlBA==",
          "signing_payload_b64": "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",
          "payload_sha256": "tBhCyLdm8wD9AdCo8KzxsZYvL05cuXS5dlNIpL4gVGI="
        },
        "build_page": "/inferences/deepseek-v3-2/devstral-small-2505-speciale-distill~teichai/teichai-iq3_xs-ollama",
        "hub_url": "https://huggingface.co/teichai/devstral-small-2505-deepseek-v3.2-speciale-distill-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "662a4e"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 0.36,
        "context_tokens": 4096,
        "build": "IQ3_XS (3-bit) by TeichAI",
        "ref": "hf.co/TeichAI/Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill-GGUF:IQ3_XS",
        "engine": "ollama-host",
        "cost_usd_per_million": 105.2094,
        "cost_breakdown": {
          "tokens_per_sec": 0.36,
          "usd_per_hour": 0.1372,
          "usd_per_million": 105.2094,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 82.829
            },
            {
              "resource": "Memory",
              "quantity": "9.2 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 7.689,
              "usd_per_million": 21.2291
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 1.1513
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD EPYC 7B12",
            "model_raw": "AMD EPYC 7B12",
            "clock_ghz": null,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 512,
            "l3_mb": 16,
            "single_core_hashes_per_sec": 1460.2,
            "multi_core_hashes_per_sec": 3246.1,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T01:38:20.212Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 6.3,
            "latency_ns": 21.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T01:38:21.135Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4982.5,
            "write_mb_s": 179.3,
            "random_read_iops": 476287,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-15T01:38:21.741Z",
            "measured_in": "TW_110",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-15T06:44:44.996Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
          "signature": "YKSegtdbk3BFhfs4/FH9yVX6NFoPLjmSlaFS0ZpkFUuDBmD4srmGn7dgiwir/KQtarmq6qTVAiZWCTonIjYlBA==",
          "signing_payload_b64": "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",
          "payload_sha256": "tBhCyLdm8wD9AdCo8KzxsZYvL05cuXS5dlNIpL4gVGI="
        },
        "build_page": "/inferences/deepseek-v3-2/devstral-small-2505-speciale-distill~teichai/teichai-iq3_xs-ollama",
        "hub_url": "https://huggingface.co/teichai/devstral-small-2505-deepseek-v3.2-speciale-distill-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "662a4e"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 0.36,
          "context_tokens": 4096,
          "build": "IQ3_XS (3-bit) by TeichAI",
          "ref": "hf.co/TeichAI/Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill-GGUF:IQ3_XS",
          "engine": "ollama-host",
          "cost_usd_per_million": 105.2094,
          "cost_breakdown": {
            "tokens_per_sec": 0.36,
            "usd_per_hour": 0.1372,
            "usd_per_million": 105.2094,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 82.829
              },
              {
                "resource": "Memory",
                "quantity": "9.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 7.689,
                "usd_per_million": 21.2291
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 1.1513
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:44:44.996Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "YKSegtdbk3BFhfs4/FH9yVX6NFoPLjmSlaFS0ZpkFUuDBmD4srmGn7dgiwir/KQtarmq6qTVAiZWCTonIjYlBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "tBhCyLdm8wD9AdCo8KzxsZYvL05cuXS5dlNIpL4gVGI="
          },
          "build_page": "/inferences/deepseek-v3-2/devstral-small-2505-speciale-distill~teichai/teichai-iq3_xs-ollama",
          "hub_url": "https://huggingface.co/teichai/devstral-small-2505-deepseek-v3.2-speciale-distill-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/TeichAI/Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "0.4 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Qwen 3 Coder-30B-A3B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3/coder-30b-a3b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct",
      "download_gb": 16,
      "params_b": 30,
      "active_params_b": 3,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "returns valid JSON",
        "long answers",
        "summarises sources"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Punjabi",
        "Tamil",
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": 131072,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-22T20:20:24.815Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/qwen-3/30b-a3b/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3-coder-30b-a3b-instruct-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-22T20:20:24.815Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-3/30b-a3b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3-coder-30b-a3b-instruct-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3-coder-30b-a3b-instruct",
        "url": "https://openrouter.ai/qwen/qwen3-coder-30b-a3b-instruct",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 0.6,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock/us-west-2",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3-Coder-30B-A3B-Instruct"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded. OpenRouter sells it at $0.60 per million output tokens."
    },
    {
      "model": "DeepSeek-OCR 2 Base",
      "made_by": "deepseek-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/deepseek-ocr-2/base",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-OCR-2",
      "download_gb": 1.8,
      "params_b": 3.2,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "reasons about code",
        "returns valid JSON",
        "long answers"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": "AMD Ryzen 7 5800X 8-Core",
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "GeForce RTX 3080 Ti",
          "gpu_count": 1,
          "vram_gb": 12,
          "host": "own machine",
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
        "machines": 1,
        "tokens_per_sec": 411.73,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by sabafallah",
        "ref": "hf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": 0.626,
        "cost_breakdown": {
          "tokens_per_sec": 411.73,
          "usd_per_hour": 0.9279,
          "usd_per_million": 0.626,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "15 of 16 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 150,
              "usd_per_million": 0.3643
            },
            {
              "resource": "Memory",
              "quantity": "1.8 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.513,
              "usd_per_million": 0.0037
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.2429
            },
            {
              "resource": "Energy",
              "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 6.25,
              "usd_per_million": 0.0152
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-10T11:20:52.995Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "4HYRG5+mdqESuBQ4ErLCPQPGiOwQDxPVtoxq6zDaTvsKyh7lrdfn21E4ptnYfxsHf9tne9OYhAsensMcYCbmBQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "77j02NRghpxfYrHhfhcXztyhdwWND1BnRDyCcC+fHdo="
        },
        "build_page": "/inferences/deepseek-ocr-2/base/sabafallah-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/sabafallah/deepseek-ocr-2-gguf",
        "weeks": [
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "cheapest": {
        "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": "AMD Ryzen 7 5800X 8-Core",
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "GeForce RTX 3080 Ti",
          "gpu_count": 1,
          "vram_gb": 12,
          "host": "own machine",
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
        "machines": 1,
        "tokens_per_sec": 411.73,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by sabafallah",
        "ref": "hf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": 0.626,
        "cost_breakdown": {
          "tokens_per_sec": 411.73,
          "usd_per_hour": 0.9279,
          "usd_per_million": 0.626,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "15 of 16 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 150,
              "usd_per_million": 0.3643
            },
            {
              "resource": "Memory",
              "quantity": "1.8 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.513,
              "usd_per_million": 0.0037
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.2429
            },
            {
              "resource": "Energy",
              "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 6.25,
              "usd_per_million": 0.0152
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-10T11:20:52.995Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "4HYRG5+mdqESuBQ4ErLCPQPGiOwQDxPVtoxq6zDaTvsKyh7lrdfn21E4ptnYfxsHf9tne9OYhAsensMcYCbmBQ==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6NDExLjczMTE1NjI1MjAyODQsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0MzA5NjUyOTk1MDMyMTAwLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwYmM2NTBiNTlhZjg3OTQ4OTYwMzFkMmU4MmJjYTBlMTE5ZjM5MGUyMWQ2OTA5NjNhZTQ3NTc4MjM4MDg0YmY1NCIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoid2luZG93cy9hbWQ2NC9udmlkaWEvMTZnIiwibW9kZWwiOiJoZi5jby9zYWJhZmFsbGFoL0RlZXBTZWVrLU9DUi0yLUdHVUY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hIiwic2FtcGxlX2NvdW50Ijo1MSwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjozNS45MjM5MjY5OTgsInRvdGFsX3Rva2VucyI6MTQ3OTF9fQ==",
          "payload_sha256": "77j02NRghpxfYrHhfhcXztyhdwWND1BnRDyCcC+fHdo="
        },
        "build_page": "/inferences/deepseek-ocr-2/base/sabafallah-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/sabafallah/deepseek-ocr-2-gguf",
        "weeks": [
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "machine_kinds": 3,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": 411.73,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by sabafallah",
          "ref": "hf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 0.626,
          "cost_breakdown": {
            "tokens_per_sec": 411.73,
            "usd_per_hour": 0.9279,
            "usd_per_million": 0.626,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.3643
              },
              {
                "resource": "Memory",
                "quantity": "1.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.513,
                "usd_per_million": 0.0037
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.2429
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0152
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "4HYRG5+mdqESuBQ4ErLCPQPGiOwQDxPVtoxq6zDaTvsKyh7lrdfn21E4ptnYfxsHf9tne9OYhAsensMcYCbmBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "77j02NRghpxfYrHhfhcXztyhdwWND1BnRDyCcC+fHdo="
          },
          "build_page": "/inferences/deepseek-ocr-2/base/sabafallah-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/sabafallah/deepseek-ocr-2-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 334.82,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by sabafallah",
          "ref": "hf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.6487,
          "cost_breakdown": {
            "tokens_per_sec": 334.82,
            "usd_per_hour": 0.7819,
            "usd_per_million": 0.6487,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 0.3285
              },
              {
                "resource": "Memory",
                "quantity": "1.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.513,
                "usd_per_million": 0.0045
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.2987
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.017
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "dnan915zASGOA/HXQ3Ttqx4niiHeRKYwoxlgAdV3bu9/RxWsak30C6bwLkw22mpr84jSJpbvh4ws6/KcHvRbCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "At5RWBQFLG5B6sgKkhEe1ty7aQJ7BhyiNckdD899ZwQ="
          },
          "build_page": "/inferences/deepseek-ocr-2/base/sabafallah-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/sabafallah/deepseek-ocr-2-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 15.36,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by sabafallah",
          "ref": "hf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 4.7185,
          "cost_breakdown": {
            "tokens_per_sec": 15.36,
            "usd_per_hour": 0.2609,
            "usd_per_million": 4.7185,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 4.5568
              },
              {
                "resource": "Memory",
                "quantity": "1.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.513,
                "usd_per_million": 0.0985
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0633
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "t3MoE3dafF90/XXpeHCShcVUq8ypYrk9B9sWbGrtUDa0hwG3KqHZSLhYWiIA0NNGawXcr5u45AOKyhJUhrAzBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "UcmVc07/VVPZHH94nQgV99MmfbBugE3CbBOHsYPY5+c="
          },
          "build_page": "/inferences/deepseek-ocr-2/base/sabafallah-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/sabafallah/deepseek-ocr-2-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-OCR-2"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "412 tokens a second on Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "TranslateGemma 4B",
      "made_by": "google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/translategemma/4b",
      "hub_url": "https://huggingface.co/google/translategemma-4b-it",
      "download_gb": 2.3,
      "params_b": 4,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 131072,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "returns valid JSON",
        "long answers"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Marathi",
        "Punjabi"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 77.54,
        "context_tokens": 131072,
        "build": "Q4_K_M (4-bit) by bullerwins",
        "ref": "hf.co/bullerwins/translategemma-4b-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.8066,
        "cost_breakdown": {
          "tokens_per_sec": 77.54,
          "usd_per_hour": 0.7835,
          "usd_per_million": 2.8066,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 1.4186
            },
            {
              "resource": "Memory",
              "quantity": "2.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.932,
              "usd_per_million": 0.0249
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 1.2896
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0734
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "cKhWjOmRDh9Gk2zmUbPpAeuHGjrYWQ2myibKuk9lMVKWOfqbwFL3iyjhQNMNUFD3DY23llJgdWht7dAMumQtBA==",
          "signing_payload_b64": "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",
          "payload_sha256": "qHZIk/D5bwnu2sLwRk+knlumQVrQ3izZO9C47Yt6GUA="
        },
        "build_page": "/inferences/translategemma/4b/bullerwins-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/bullerwins/translategemma-4b-it-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 77.54,
        "context_tokens": 131072,
        "build": "Q4_K_M (4-bit) by bullerwins",
        "ref": "hf.co/bullerwins/translategemma-4b-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.8066,
        "cost_breakdown": {
          "tokens_per_sec": 77.54,
          "usd_per_hour": 0.7835,
          "usd_per_million": 2.8066,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 1.4186
            },
            {
              "resource": "Memory",
              "quantity": "2.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.932,
              "usd_per_million": 0.0249
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 1.2896
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0734
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "cKhWjOmRDh9Gk2zmUbPpAeuHGjrYWQ2myibKuk9lMVKWOfqbwFL3iyjhQNMNUFD3DY23llJgdWht7dAMumQtBA==",
          "signing_payload_b64": "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",
          "payload_sha256": "qHZIk/D5bwnu2sLwRk+knlumQVrQ3izZO9C47Yt6GUA="
        },
        "build_page": "/inferences/translategemma/4b/bullerwins-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/bullerwins/translategemma-4b-it-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 77.54,
          "context_tokens": 131072,
          "build": "Q4_K_M (4-bit) by bullerwins",
          "ref": "hf.co/bullerwins/translategemma-4b-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.8066,
          "cost_breakdown": {
            "tokens_per_sec": 77.54,
            "usd_per_hour": 0.7835,
            "usd_per_million": 2.8066,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 1.4186
              },
              {
                "resource": "Memory",
                "quantity": "2.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.932,
                "usd_per_million": 0.0249
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.2896
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0734
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "cKhWjOmRDh9Gk2zmUbPpAeuHGjrYWQ2myibKuk9lMVKWOfqbwFL3iyjhQNMNUFD3DY23llJgdWht7dAMumQtBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "qHZIk/D5bwnu2sLwRk+knlumQVrQ3izZO9C47Yt6GUA="
          },
          "build_page": "/inferences/translategemma/4b/bullerwins-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bullerwins/translategemma-4b-it-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Gemma terms of use",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/google/translategemma-4b-it"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "78 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
      "made_by": "mingxilei",
      "maker_logo": "https://huggingface.co/avatars/f750d88079fbb685f3229b20bac341bd.svg",
      "based_on": "FunctionGemma 270M",
      "variant_of": "/inferences/functiongemma/270m",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat",
      "hub_url": "https://huggingface.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
      "download_gb": 0.2,
      "params_b": 0.27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "reasons about code",
        "returns valid JSON"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 39.3,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by mingxilei",
        "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.779,
        "cost_breakdown": {
          "tokens_per_sec": 39.3,
          "usd_per_hour": 0.1102,
          "usd_per_million": 0.779,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.7634
            },
            {
              "resource": "Memory",
              "quantity": "0.2 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.196,
              "usd_per_million": 0.005
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0106
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-25T07:41:14.640Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
          "signature": "X0c/TQdxAjPDBm5R//2Zbl/6sJrak/DTsQ7XkKjiH/cEFfgx2tGndoGR30Ae/jI1bntQ0mgWphCeCYwcTQoOAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "lABtwKNxFNzg0EYUCvyDy1MjMKnmpdtRxElG0SOOqr8="
        },
        "build_page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat/mingxilei-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-qat-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "acbe75"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 39.3,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by mingxilei",
        "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.779,
        "cost_breakdown": {
          "tokens_per_sec": 39.3,
          "usd_per_hour": 0.1102,
          "usd_per_million": 0.779,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.7634
            },
            {
              "resource": "Memory",
              "quantity": "0.2 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.196,
              "usd_per_million": 0.005
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0106
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-25T07:41:14.640Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
          "signature": "X0c/TQdxAjPDBm5R//2Zbl/6sJrak/DTsQ7XkKjiH/cEFfgx2tGndoGR30Ae/jI1bntQ0mgWphCeCYwcTQoOAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "lABtwKNxFNzg0EYUCvyDy1MjMKnmpdtRxElG0SOOqr8="
        },
        "build_page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat/mingxilei-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-qat-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "acbe75"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 4,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 39.3,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by mingxilei",
          "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.779,
          "cost_breakdown": {
            "tokens_per_sec": 39.3,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.779,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.7634
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.005
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0106
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "X0c/TQdxAjPDBm5R//2Zbl/6sJrak/DTsQ7XkKjiH/cEFfgx2tGndoGR30Ae/jI1bntQ0mgWphCeCYwcTQoOAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "lABtwKNxFNzg0EYUCvyDy1MjMKnmpdtRxElG0SOOqr8="
          },
          "build_page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat/mingxilei-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-qat-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 31.81,
          "context_tokens": null,
          "build": "Q5_K_M (5-bit) by mingxilei",
          "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q5_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9624,
          "cost_breakdown": {
            "tokens_per_sec": 31.81,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.9624,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.943
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.201,
                "usd_per_million": 0.0063
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0131
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "d0z8CazX/Uw7H/Zr4kTP3hYof5QIOukPUxkjycUu1th2pnVI4aAjBA99ceF611iDgpktUt1P00wcCT6/ebEWDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "YTXsNroHaiXX71xymeYQoWUggSCD5QcqUGzIQWW3AI4="
          },
          "build_page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat/mingxilei-q5_k_m-ollama",
          "hub_url": "https://huggingface.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-qat-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 31.48,
          "context_tokens": null,
          "build": "Q8_0 (8-bit) by mingxilei",
          "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9735,
          "cost_breakdown": {
            "tokens_per_sec": 31.48,
            "usd_per_hour": 0.1103,
            "usd_per_million": 0.9735,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.9531
              },
              {
                "resource": "Memory",
                "quantity": "0.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.226,
                "usd_per_million": 0.0072
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0132
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "ZPxeA4Ox8tjWTaC2+qyBFHznn8oe/gl3FGT1FQGE4ON25ak71GUKuYBXooawqMlL0Pt/RKUTEU4Wx04zznVgAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "YfmBkQrAZePfjUBqozpBtrntyhkfk/x8A4i5KrD2hGY="
          },
          "build_page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat/mingxilei-q8_0-ollama",
          "hub_url": "https://huggingface.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-qat-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 23.13,
          "context_tokens": null,
          "build": "F16 (16-bit, full precision) by mingxilei",
          "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:F16",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.3335,
          "cost_breakdown": {
            "tokens_per_sec": 23.13,
            "usd_per_hour": 0.111,
            "usd_per_million": 1.3335,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.2973
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.421,
                "usd_per_million": 0.0182
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.018
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "NqgdnMzZSS9xU892U8djThcSq1GosDCGZjuQgLv3pn19JAq2u+GXKJOVClpH8Fc1RjemXFM0086SU+62g/qpDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "l4jWnAmGi5Of+pax/Pcyp1gkBGlMzXP/0DnipKcLVrU="
          },
          "build_page": "/inferences/functiongemma/270m~mingxilei-unsloth-ft-qat/mingxilei-f16-ollama",
          "hub_url": "https://huggingface.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-qat-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": "https://huggingface.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF"
      },
      "builds": 4,
      "variants": [],
      "bottom_line": "39 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Gemma 3 1B",
      "made_by": "google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/gemma-3/1b",
      "hub_url": "https://huggingface.co/google/gemma-3-1b-it",
      "download_gb": 0.8,
      "params_b": 1,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 8192,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "returns valid JSON",
        "long answers"
      ],
      "languages_passed": [
        "English",
        "Gujarati",
        "Hindi",
        "Marathi",
        "Dutch",
        "Russian",
        "Telugu",
        "Bengali"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 2,
        "tokens_per_sec": 13.98,
        "context_tokens": 8192,
        "build": "Q4_K_M (4-bit) by ggml-org",
        "ref": "hf.co/ggml-org/gemma-3-1b-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.2199,
        "cost_breakdown": {
          "tokens_per_sec": 13.98,
          "usd_per_hour": 0.1118,
          "usd_per_million": 2.2199,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 2.1454
            },
            {
              "resource": "Memory",
              "quantity": "0.8 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.625,
              "usd_per_million": 0.0447
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0298
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "yP8sNk1aVG8ndDXE3hGWT1Lp8lzjKWxYX6FjqZf2L7l06/ZqPmc5po2O5nITnjxVhoBkid68D+gbE8zS0zxDCA==",
          "signing_payload_b64": "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",
          "payload_sha256": "SGmJiRvuTNS7M/OJOAxNYjDLOspJvqZkROzm3E8XXD8="
        },
        "build_page": "/inferences/gemma-3/1b/ggml-org-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/ggml-org/gemma-3-1b-it-gguf",
        "weeks": [
          "2026-W43",
          "2026-W44"
        ],
        "providers": [
          "e61d3b",
          "acbe75"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 2,
        "tokens_per_sec": 13.98,
        "context_tokens": 8192,
        "build": "Q4_K_M (4-bit) by ggml-org",
        "ref": "hf.co/ggml-org/gemma-3-1b-it-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 2.2199,
        "cost_breakdown": {
          "tokens_per_sec": 13.98,
          "usd_per_hour": 0.1118,
          "usd_per_million": 2.2199,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 2.1454
            },
            {
              "resource": "Memory",
              "quantity": "0.8 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.625,
              "usd_per_million": 0.0447
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0298
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "yP8sNk1aVG8ndDXE3hGWT1Lp8lzjKWxYX6FjqZf2L7l06/ZqPmc5po2O5nITnjxVhoBkid68D+gbE8zS0zxDCA==",
          "signing_payload_b64": "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",
          "payload_sha256": "SGmJiRvuTNS7M/OJOAxNYjDLOspJvqZkROzm3E8XXD8="
        },
        "build_page": "/inferences/gemma-3/1b/ggml-org-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/ggml-org/gemma-3-1b-it-gguf",
        "weeks": [
          "2026-W43",
          "2026-W44"
        ],
        "providers": [
          "e61d3b",
          "acbe75"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 7,
      "weeks_measured": [
        "2026-W43",
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 2,
          "tokens_per_sec": 13.98,
          "context_tokens": 8192,
          "build": "Q4_K_M (4-bit) by ggml-org",
          "ref": "hf.co/ggml-org/gemma-3-1b-it-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.2199,
          "cost_breakdown": {
            "tokens_per_sec": 13.98,
            "usd_per_hour": 0.1118,
            "usd_per_million": 2.2199,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.1454
              },
              {
                "resource": "Memory",
                "quantity": "0.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.625,
                "usd_per_million": 0.0447
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0298
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "yP8sNk1aVG8ndDXE3hGWT1Lp8lzjKWxYX6FjqZf2L7l06/ZqPmc5po2O5nITnjxVhoBkid68D+gbE8zS0zxDCA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTMuOTgzMjQxNDkxMzEzNjE5LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5MzQwNTkxNzI5ODY0ODg5MSwibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMGVlMzVkYzQ2YTIyMTJjYjMxNDQ0Y2E0YzE5NjY5YWM4ZDk3YjA5MjUyMGI4YzllNGQyOWFiNDhkOWVlNjFkM2IiLCJleHRyYSI6eyJoYXJkd2FyZV9jbGFzcyI6ImxpbnV4L2FtZDY0L25vbmUvMzJnIiwibW9kZWwiOiJoZi5jby9nZ21sLW9yZy9nZW1tYS0zLTFiLWl0LUdHVUY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjQ4LCJzb3VyY2UiOiJtZXRyaWNzLm1vZGVsc190b2tlbnNfcGVyX3NlYyIsInRvdGFsX3NlY29uZHMiOjE2My42OTU5NTAwMDAwMDAwNCwidG90YWxfdG9rZW5zIjoyMjg5fX0=",
            "payload_sha256": "SGmJiRvuTNS7M/OJOAxNYjDLOspJvqZkROzm3E8XXD8="
          },
          "build_page": "/inferences/gemma-3/1b/ggml-org-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/ggml-org/gemma-3-1b-it-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44"
          ],
          "providers": [
            "e61d3b",
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 13.61,
          "context_tokens": 8192,
          "build": "Q3_K_S (3-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/gemma-3-1b-it-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.2346,
          "cost_breakdown": {
            "tokens_per_sec": 13.61,
            "usd_per_hour": 0.1095,
            "usd_per_million": 2.2346,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.204
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0306
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "7yl4afdSdOMaQEsDrFBItwCgR4GCyYGr67MTFtbDE7buifN3RQkIckux4hO+LYjfWSNcKqK3cZ6H8j3YykImDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "mmew07fzANs2QicBDyy8uRTn+q/ZdJeIAZONj+ZMVVs="
          },
          "build_page": "/inferences/gemma-3/1b/maziyarpanahi-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/gemma-3-1b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 13.1,
          "context_tokens": 8192,
          "build": "Q3_K_L (3-bit) by lmstudio-community",
          "ref": "hf.co/lmstudio-community/gemma-3-1b-it-GGUF:Q3_K_L",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.3671,
          "cost_breakdown": {
            "tokens_per_sec": 13.1,
            "usd_per_hour": 0.1116,
            "usd_per_million": 2.3671,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.2908
              },
              {
                "resource": "Memory",
                "quantity": "0.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.583,
                "usd_per_million": 0.0445
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0318
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "7spjQg07k42DaGzFulWHl1FioLZLS45sDztgcXDq9tRDL4EDDbW1nG/Hmt1ESeW22ExSVhsuHhVsqiimgcXOCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "8oj2rt9Vxwh9EaLvJ8VagyX9vARLb8a0XYzFdubYuww="
          },
          "build_page": "/inferences/gemma-3/1b/lmstudio-community-q3_k_l-ollama",
          "hub_url": "https://huggingface.co/lmstudio-community/gemma-3-1b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 12.96,
          "context_tokens": 8192,
          "build": "Q2_K (2-bit) by maziyarpanahi",
          "ref": "hf.co/MaziyarPanahi/gemma-3-1b-it-GGUF:Q2_K",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.3461,
          "cost_breakdown": {
            "tokens_per_sec": 12.96,
            "usd_per_hour": 0.1095,
            "usd_per_million": 2.3461,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.314
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0322
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "kw84W7bw8C7vH87eJ9SZjXef38Xu67Cz+eVr65sYi9pM7WSJcJc4zb7pshzl4/tdsRbaHSSsWEOJdoC+t01KBQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "XoFBbY0HlHb4doKHQf03LXkNUgTK3EhUmmd6pWZjYFw="
          },
          "build_page": "/inferences/gemma-3/1b/maziyarpanahi-q2_k-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/gemma-3-1b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 10.79,
          "context_tokens": 8192,
          "build": "Q6_K (6-bit) by lmstudio-community",
          "ref": "hf.co/lmstudio-community/gemma-3-1b-it-GGUF:Q6_K",
          "engine": "ollama-host",
          "cost_usd_per_million": 2.8924,
          "cost_breakdown": {
            "tokens_per_sec": 10.79,
            "usd_per_hour": 0.1123,
            "usd_per_million": 2.8924,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 2.781
              },
              {
                "resource": "Memory",
                "quantity": "0.9 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.785,
                "usd_per_million": 0.0728
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0387
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "s91a8yr+9Xwwje9YtTK4zQs7c/X0+3YUQvCjx+FEPlZOmDsmQucLtsx5yQIPGBDAve9itNYY0ram5h/NN75+AA==",
            "signing_payload_b64": "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",
            "payload_sha256": "WvQXoojc7FNivdkVG+IK6CsiGLwRfoIR14+nJYgSwrk="
          },
          "build_page": "/inferences/gemma-3/1b/lmstudio-community-q6_k-ollama",
          "hub_url": "https://huggingface.co/lmstudio-community/gemma-3-1b-it-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 9.53,
          "context_tokens": 8192,
          "build": "Q8_0 (8-bit) by ggml-org",
          "ref": "hf.co/ggml-org/gemma-3-1b-it-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 3.2791,
          "cost_breakdown": {
            "tokens_per_sec": 9.53,
            "usd_per_hour": 0.1125,
            "usd_per_million": 3.2791,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 3.1483
              },
              {
                "resource": "Memory",
                "quantity": "1.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.829,
                "usd_per_million": 0.087
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0438
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "VThqWMXZ43YMZ9HbSf/Eg9DqFmQapYGT8W2ZiunK7qyWu01meHtxMao9XGVIswa8F8rX7cYuzaV5NOZDrnMyCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "jNol40QIbdH55GXg0kkQWmcnsv5S4lW1KdlXb/ZjnyE="
          },
          "build_page": "/inferences/gemma-3/1b/ggml-org-q8_0-ollama",
          "hub_url": "https://huggingface.co/ggml-org/gemma-3-1b-it-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Gemma terms of use",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/google/gemma-3-1b-it"
      },
      "builds": 6,
      "variants": [],
      "bottom_line": "14 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "DeepSeek-OCR Base",
      "made_by": "deepseek-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/deepseek-ocr/base",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-OCR",
      "download_gb": 1.8,
      "params_b": 3.2,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "returns valid JSON",
        "summarises sources"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 340.22,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by sabafallah",
        "ref": "hf.co/sabafallah/DeepSeek-OCR-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.6384,
        "cost_breakdown": {
          "tokens_per_sec": 340.22,
          "usd_per_hour": 0.7819,
          "usd_per_million": 0.6384,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 0.3233
            },
            {
              "resource": "Memory",
              "quantity": "1.8 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.513,
              "usd_per_million": 0.0044
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.2939
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0167
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "MldOjzjP4hwZYEELAoIg6JxlNEbRpD5Jl76MA5awvmFEE1hThq8QYNnXKQkzyGga54Xr+8vBwD+lps1pWwABCQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "jHQ1GDVSNrtTn5osqOMTcXALb6OfiZocbGwYFpH6n2w="
        },
        "build_page": "/inferences/deepseek-ocr/base/sabafallah-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/sabafallah/deepseek-ocr-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 340.22,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by sabafallah",
        "ref": "hf.co/sabafallah/DeepSeek-OCR-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.6384,
        "cost_breakdown": {
          "tokens_per_sec": 340.22,
          "usd_per_hour": 0.7819,
          "usd_per_million": 0.6384,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 0.3233
            },
            {
              "resource": "Memory",
              "quantity": "1.8 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.513,
              "usd_per_million": 0.0044
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.2939
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0167
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "MldOjzjP4hwZYEELAoIg6JxlNEbRpD5Jl76MA5awvmFEE1hThq8QYNnXKQkzyGga54Xr+8vBwD+lps1pWwABCQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "jHQ1GDVSNrtTn5osqOMTcXALb6OfiZocbGwYFpH6n2w="
        },
        "build_page": "/inferences/deepseek-ocr/base/sabafallah-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/sabafallah/deepseek-ocr-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 2,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 340.22,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by sabafallah",
          "ref": "hf.co/sabafallah/DeepSeek-OCR-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.6384,
          "cost_breakdown": {
            "tokens_per_sec": 340.22,
            "usd_per_hour": 0.7819,
            "usd_per_million": 0.6384,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 0.3233
              },
              {
                "resource": "Memory",
                "quantity": "1.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.513,
                "usd_per_million": 0.0044
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.2939
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0167
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "MldOjzjP4hwZYEELAoIg6JxlNEbRpD5Jl76MA5awvmFEE1hThq8QYNnXKQkzyGga54Xr+8vBwD+lps1pWwABCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "jHQ1GDVSNrtTn5osqOMTcXALb6OfiZocbGwYFpH6n2w="
          },
          "build_page": "/inferences/deepseek-ocr/base/sabafallah-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/sabafallah/deepseek-ocr-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 13.32,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by sabafallah",
          "ref": "hf.co/sabafallah/DeepSeek-OCR-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 5.4435,
          "cost_breakdown": {
            "tokens_per_sec": 13.32,
            "usd_per_hour": 0.2609,
            "usd_per_million": 5.4435,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 5.2569
              },
              {
                "resource": "Memory",
                "quantity": "1.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.513,
                "usd_per_million": 0.1136
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.073
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "fCcoLkBF4l/BECmDHM40HrCL2mPJpkNSdr1QYDyv4WzGJhAcA1kPBvHOg21NFNWy2Uycu3DRh/BhZ9ku0vt3Bg==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MTMuMzE1OTM4Nzk5MDg5MTMsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0NzUxNDQ2ODgyNTg0OTAwLCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwMTUzNmRmOTRmODhjMmYyZGNlOWM4NzI5N2Y0MmUwOTAyZDVkNjM1ZWEyNzM2Zjk5MTQxYzNlYmNlYTJlNWMwMiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoid2luZG93cy9hbWQ2NC9pbnRlbC8xNmciLCJtb2RlbCI6ImhmLmNvL3NhYmFmYWxsYWgvRGVlcFNlZWstT0NSLUdHVUY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hIiwic2FtcGxlX2NvdW50Ijo1MSwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjo1NjkuOTkzNjA4LCJ0b3RhbF90b2tlbnMiOjc1OTB9fQ==",
            "payload_sha256": "2wwrJBEw+e8137zc3dLRhDSekaQbSJZ59qFU7z2ksxM="
          },
          "build_page": "/inferences/deepseek-ocr/base/sabafallah-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/sabafallah/deepseek-ocr-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "2e5c02"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-OCR"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "340 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Sarvam-1",
      "made_by": "sarvamai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/60270a7c32856987162c641a/umd13GCWVijwTDGZzw3q-.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/sarvam/1",
      "hub_url": "https://huggingface.co/sarvamai/sarvam-1",
      "download_gb": 1.4,
      "params_b": 2,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 8192,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "returns valid JSON",
        "long answers"
      ],
      "languages_passed": [
        "Bengali",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Marathi",
        "Punjabi",
        "Telugu",
        "Malayalam",
        "Tamil"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 121.44,
        "context_tokens": 8192,
        "build": "Q4_K_M (4-bit) by bartowski",
        "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 1.7861,
        "cost_breakdown": {
          "tokens_per_sec": 121.44,
          "usd_per_hour": 0.7808,
          "usd_per_million": 1.7861,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 0.9058
            },
            {
              "resource": "Memory",
              "quantity": "1.4 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.201,
              "usd_per_million": 0.0099
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.8235
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.0469
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "RDQRzBO3Ke2zwZxG59t3qpuu4vt+0iyh+KT5WvNfvzFgNA7u9czHmPcg+QXP9UQD2DuW4s9XkZB1Olzfa3cuCg==",
          "signing_payload_b64": "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",
          "payload_sha256": "kKojnbs6fubM5IETsHxYSbTKvI0PpXNM+4MHaHbzCDI="
        },
        "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/bartowski/sarvam-1-gguf",
        "weeks": [
          "2026-W37",
          "2026-W38",
          "2026-W42",
          "2026-W43",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": 111.22,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by bartowski",
        "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 1.2208,
        "cost_breakdown": {
          "tokens_per_sec": 111.22,
          "usd_per_hour": 0.4888,
          "usd_per_million": 1.2208,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.2697
            },
            {
              "resource": "Memory",
              "quantity": "1.4 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 1.201,
              "usd_per_million": 0.0108
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.8991
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.0412
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "N8h+/CDdHI+jzwc8eWutheTSY09LHF7CXfHHdhkLHLWL7/oUNzEAKT2OcdnCa978pwaik/R1kFY0R26nTNBzDQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "iFRMIC2eIdiwVAww3vrDwSrTUAaSSpE2Nj0vL7BXhpQ="
        },
        "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/bartowski/sarvam-1-gguf",
        "weeks": [
          "2026-W37",
          "2026-W38",
          "2026-W42",
          "2026-W43",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "machine_kinds": 7,
      "machines_total": 8,
      "weeks_measured": [
        "2026-W37",
        "2026-W38",
        "2026-W42",
        "2026-W43",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 121.44,
          "context_tokens": 8192,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.7861,
          "cost_breakdown": {
            "tokens_per_sec": 121.44,
            "usd_per_hour": 0.7808,
            "usd_per_million": 1.7861,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 0.9058
              },
              {
                "resource": "Memory",
                "quantity": "1.4 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.201,
                "usd_per_million": 0.0099
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.8235
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.0469
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "RDQRzBO3Ke2zwZxG59t3qpuu4vt+0iyh+KT5WvNfvzFgNA7u9czHmPcg+QXP9UQD2DuW4s9XkZB1Olzfa3cuCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "kKojnbs6fubM5IETsHxYSbTKvI0PpXNM+4MHaHbzCDI="
          },
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/sarvam-1-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W42",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 111.22,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.2208,
          "cost_breakdown": {
            "tokens_per_sec": 111.22,
            "usd_per_hour": 0.4888,
            "usd_per_million": 1.2208,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.2697
              },
              {
                "resource": "Memory",
                "quantity": "1.4 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.201,
                "usd_per_million": 0.0108
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.8991
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0412
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "N8h+/CDdHI+jzwc8eWutheTSY09LHF7CXfHHdhkLHLWL7/oUNzEAKT2OcdnCa978pwaik/R1kFY0R26nTNBzDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "iFRMIC2eIdiwVAww3vrDwSrTUAaSSpE2Nj0vL7BXhpQ="
          },
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/sarvam-1-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W42",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": 62.17,
          "context_tokens": 8192,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 3.1624,
          "cost_breakdown": {
            "tokens_per_sec": 62.17,
            "usd_per_hour": 0.7078,
            "usd_per_million": 3.1624,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "9 of 10 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 90,
                "usd_per_million": 1.4476
              },
              {
                "resource": "Memory",
                "quantity": "1.4 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.201,
                "usd_per_million": 0.0193
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 1.6084
              },
              {
                "resource": "Energy",
                "quantity": "195 W modelled (9 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.417,
                "usd_per_million": 0.0871
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-02T11:31:50.442Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ad2ba1fabbc8e5f23892c0134225790dd035d652ce541dc16e0b098c56afbb8a",
            "signature": "7Mzrnr6rMNlpSm5ORXmlpggnh+MORaGzOqNxSLN1c1pdI/9dI4FTjkkaE3GENmfI16MVzddjWopPHWH0IDYmCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "0mjk6XLJMNmeA5rkOnGSVK8GLYfGbWRqcdgcSRO8xkQ="
          },
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/sarvam-1-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W42",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Windows · NVIDIA 4 GB · 8 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 4 GB · 8 CPUs · 8 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 4,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|8|8|nvidia|4|||"
          },
          "machine_class": "windows|amd64|8|8|nvidia|4|||",
          "machines": 1,
          "tokens_per_sec": 23.48,
          "context_tokens": 8192,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 7.5107,
          "cost_breakdown": {
            "tokens_per_sec": 23.48,
            "usd_per_hour": 0.6348,
            "usd_per_million": 7.5107,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 2.9814
              },
              {
                "resource": "Memory",
                "quantity": "1.4 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.201,
                "usd_per_million": 0.0512
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 4.2592
              },
              {
                "resource": "Energy",
                "quantity": "185 W modelled (7 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.139,
                "usd_per_million": 0.2189
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-26T10:45:15.368Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220df04c29e6a0ad2ef0dc554f22fdbb03346428fde27cc62255d9dd9c61412954d",
            "signature": "XQOANHZUv5uZTkhIKpy3FUmfU2Fpan+cDJEruzYZSL21nNB8Pq06FL4PXreMRE3oXDqN17IOJKpdRt69sOtcBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "1nhtmpsdU4jH2I1tMjpeMpZXlxoMi9hCRX4Cq4pe8yQ="
          },
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/sarvam-1-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W42",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "12954d"
          ]
        },
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": 19.76,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 13.0304,
          "cost_breakdown": {
            "tokens_per_sec": 19.76,
            "usd_per_hour": 0.9268,
            "usd_per_million": 13.0304,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 7.592
              },
              {
                "resource": "Memory",
                "quantity": "1.4 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.201,
                "usd_per_million": 0.0608
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 5.0613
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.3163
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "F5Jbiivz/n2bnD4rajHRfruggRLe1yf+AS4X33N+kCSsurC0wAHLRvFgacwSWdSb4tIVK1hSp1jeA4zh2k6QCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "xr717MF2lwiKoiUdiQo17Th06gp/cYr9zTA0dstiJc8="
          },
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/sarvam-1-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W42",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": 8.1,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 8.9138,
          "cost_breakdown": {
            "tokens_per_sec": 8.1,
            "usd_per_hour": 0.2598,
            "usd_per_million": 8.9138,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 8.6455
              },
              {
                "resource": "Memory",
                "quantity": "1.4 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.201,
                "usd_per_million": 0.1483
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.12
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "4AF9srIpWtrtXuqApmRSZDoZwf5KezUUcayXypSU4T5NTfUInkHA6t9baB2CQh1wAXvq/4LgCm+oG8S+QUCSBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "8Y5eP0X7r/hzHXpQuDYRvLz+ATLb03Ok2l3VDq0BRdQ="
          },
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/sarvam-1-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W42",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 4.83,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by bartowski",
          "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 14.9543,
          "cost_breakdown": {
            "tokens_per_sec": 4.83,
            "usd_per_hour": 0.2598,
            "usd_per_million": 14.9543,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 14.504
              },
              {
                "resource": "Memory",
                "quantity": "1.4 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 1.201,
                "usd_per_million": 0.2488
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.2014
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "zOzcOojwFSJa7iY6P5I80gNY8jssi+BFdjn17lLOoOPsjgao35PnumPcrLTVuE3LxO1137PaHWvuL5NKQbygCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "c3rFbcnYJtjbMbmxSYoj1nDeHBECdXtsM+T8bXwVrOM="
          },
          "build_page": "/inferences/sarvam/1/bartowski-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/bartowski/sarvam-1-gguf",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W42",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "BF16 (16-bit, full precision) by sarvamai",
          "ref": "sarvamai/sarvam-1",
          "engine": "vllm",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-20T09:39:55.520Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sarvam/1/sarvamai-bf16-vllm",
          "hub_url": "https://huggingface.co/sarvamai/sarvam-1",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": "https://huggingface.co/sarvamai/sarvam-1"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "121 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "functiongemma-270m-it-abominable-heresy",
      "made_by": "MuXodious",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/65d0f78d98b4a470bf2aca80/lSw3a-hajldtu2JnlptOr.png",
      "based_on": "FunctionGemma 270M",
      "variant_of": "/inferences/functiongemma/270m",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/functiongemma/270m~muxodious-abominable-heresy",
      "hub_url": "https://huggingface.co/MuXodious/functiongemma-270m-it-abominable-heresy",
      "download_gb": 0.2,
      "params_b": 0.27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "writes code",
        "returns valid JSON"
      ],
      "languages_passed": [
        "Hindi",
        "Bengali",
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 38.58,
        "context_tokens": null,
        "build": "IQ4_XS (4-bit) by mradermacher",
        "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:IQ4_XS",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.7932,
        "cost_breakdown": {
          "tokens_per_sec": 38.58,
          "usd_per_hour": 0.1102,
          "usd_per_million": 0.7932,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.7775
            },
            {
              "resource": "Memory",
              "quantity": "0.2 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.187,
              "usd_per_million": 0.0048
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0108
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-25T07:41:14.640Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
          "signature": "VehGAESlKyA6SWp7bwkmA1SQy/4QDonX/oMf0e7u02XZoyAjE4B0hedi/VKTcb9MFH4iN9lGZ7VIGVlpyRkfCw==",
          "signing_payload_b64": "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",
          "payload_sha256": "z31lTggVwG3jLbJV0JKv0sLn9srxyzn96tpRRvmUsDM="
        },
        "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-iq4_xs-ollama",
        "hub_url": "https://huggingface.co/mradermacher/functiongemma-270m-it-abominable-heresy-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "acbe75"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 38.58,
        "context_tokens": null,
        "build": "IQ4_XS (4-bit) by mradermacher",
        "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:IQ4_XS",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.7932,
        "cost_breakdown": {
          "tokens_per_sec": 38.58,
          "usd_per_hour": 0.1102,
          "usd_per_million": 0.7932,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.7775
            },
            {
              "resource": "Memory",
              "quantity": "0.2 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.187,
              "usd_per_million": 0.0048
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0108
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-25T07:41:14.640Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
          "signature": "VehGAESlKyA6SWp7bwkmA1SQy/4QDonX/oMf0e7u02XZoyAjE4B0hedi/VKTcb9MFH4iN9lGZ7VIGVlpyRkfCw==",
          "signing_payload_b64": "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",
          "payload_sha256": "z31lTggVwG3jLbJV0JKv0sLn9srxyzn96tpRRvmUsDM="
        },
        "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-iq4_xs-ollama",
        "hub_url": "https://huggingface.co/mradermacher/functiongemma-270m-it-abominable-heresy-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "acbe75"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 7,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 38.58,
          "context_tokens": null,
          "build": "IQ4_XS (4-bit) by mradermacher",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:IQ4_XS",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.7932,
          "cost_breakdown": {
            "tokens_per_sec": 38.58,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.7932,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.7775
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.187,
                "usd_per_million": 0.0048
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0108
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "VehGAESlKyA6SWp7bwkmA1SQy/4QDonX/oMf0e7u02XZoyAjE4B0hedi/VKTcb9MFH4iN9lGZ7VIGVlpyRkfCw==",
            "signing_payload_b64": "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",
            "payload_sha256": "z31lTggVwG3jLbJV0JKv0sLn9srxyzn96tpRRvmUsDM="
          },
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-iq4_xs-ollama",
          "hub_url": "https://huggingface.co/mradermacher/functiongemma-270m-it-abominable-heresy-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 36.72,
          "context_tokens": null,
          "build": "Q3_K_S (3-bit) by mradermacher",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.8333,
          "cost_breakdown": {
            "tokens_per_sec": 36.72,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.8333,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8169
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.183,
                "usd_per_million": 0.005
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0114
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "plpjGiafaRVNMDh0zQtM9KRUvvKd8W9e3xQH0/mYOBl9m/ApvqzPWVagy1Umc6wY7WVMLCng4icGIVcOIxe5CQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "TeczI9y63ptv5SjkZPefNFL0r0w+XF+jMQ1kKpGx3RA="
          },
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/mradermacher/functiongemma-270m-it-abominable-heresy-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 35.11,
          "context_tokens": null,
          "build": "Q2_K (2-bit) by mradermacher",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q2_K",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.8715,
          "cost_breakdown": {
            "tokens_per_sec": 35.11,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.8715,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8544
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.184,
                "usd_per_million": 0.0052
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0119
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "UQ+ol2lRsaufmBqgAO4vM7JFpGkQm/MgpZ/jo6qnpKEBata/5s7767LwoWf2yxr4aczYHJK8WqMC7RVb2PNGCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "iFQOgk3Bh/r3va5uy25ceka3dWnB+/eOtvnDHMu0yms="
          },
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-q2_k-ollama",
          "hub_url": "https://huggingface.co/mradermacher/functiongemma-270m-it-abominable-heresy-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 33.93,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by mradermacher",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9024,
          "cost_breakdown": {
            "tokens_per_sec": 33.93,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.9024,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8843
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.0058
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0123
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "Q5Snaad4r/moyQqiT8Hv8fy7KjrckFuZKmNFebyzVQs+UkOIFlVpE3dUQI9guNNb7qadZdwYOrwBifMsMrL/Cg==",
            "signing_payload_b64": "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",
            "payload_sha256": "jDusQ0/LL6wRXU1izuQQIcFC/asDVW6PFSLt/XmHCBA="
          },
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/mradermacher/functiongemma-270m-it-abominable-heresy-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 33.74,
          "context_tokens": null,
          "build": "Q4_K_S (4-bit) by mradermacher",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q4_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9073,
          "cost_breakdown": {
            "tokens_per_sec": 33.74,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.9073,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8892
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.194,
                "usd_per_million": 0.0058
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0124
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "EDvxRWQIUV2cpvBnp2JGI2p6osMxeJPLb7ahxUWzwxWNY2nlKmhg5gyOV9lg3o0kjk81VrfYJaLrlXw/WY1ECQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "W462zWF6heg3huc5yl3WZ8B/EWJRuVJXTeEJZiRvETY="
          },
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-q4_k_s-ollama",
          "hub_url": "https://huggingface.co/mradermacher/functiongemma-270m-it-abominable-heresy-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 32.19,
          "context_tokens": null,
          "build": "Q8_0 (8-bit) by mradermacher",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9519,
          "cost_breakdown": {
            "tokens_per_sec": 32.19,
            "usd_per_hour": 0.1103,
            "usd_per_million": 0.9519,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.9319
              },
              {
                "resource": "Memory",
                "quantity": "0.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.226,
                "usd_per_million": 0.007
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.013
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "5MzpNlm50+knSNKTzQ3QQgwa6cZwym/+YeudrZ44LylQEhzvEohIxsbkIEg1GzHNVBLXNDTpe0cwY7zgCpUeDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "SGMtarQiteaRDuDgR2K+Ei//6+tvE7g22klIabjDM2Q="
          },
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-q8_0-ollama",
          "hub_url": "https://huggingface.co/mradermacher/functiongemma-270m-it-abominable-heresy-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 22.71,
          "context_tokens": null,
          "build": "F16 (16-bit, full precision) by mradermacher",
          "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:F16",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.3578,
          "cost_breakdown": {
            "tokens_per_sec": 22.71,
            "usd_per_hour": 0.111,
            "usd_per_million": 1.3578,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.3209
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.421,
                "usd_per_million": 0.0185
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0184
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "LKKAt3FY3XjwAbqzTWeFaCVnF9iGvxaFurGkfPcYxVDWEcBlDCh78UXPhs3EhzwWY2rfo44DHxGZQhNI0LRMBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "qSbHUDWpDPf/fN8j6sdSpvFeY/ZYgx0/S9oHUz0NSAo="
          },
          "build_page": "/inferences/functiongemma/270m~muxodious-abominable-heresy/mradermacher-f16-ollama",
          "hub_url": "https://huggingface.co/mradermacher/functiongemma-270m-it-abominable-heresy-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Gemma terms of use",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/MuXodious/functiongemma-270m-it-abominable-heresy"
      },
      "builds": 7,
      "variants": [],
      "bottom_line": "39 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Airavata 7B (Hindi)",
      "made_by": "ai4bharat",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1645373988267-5fbfc7a6e366524fe8e97ccf.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/airavata/7b",
      "hub_url": "https://huggingface.co/ai4bharat/Airavata",
      "download_gb": 3.9,
      "params_b": 7,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": 4096,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "reasons about code",
        "long answers"
      ],
      "languages_passed": [
        "English",
        "Hindi"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": 31.59,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by mradermacher",
        "ref": "hf.co/mradermacher/Airavata-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 4.3623,
        "cost_breakdown": {
          "tokens_per_sec": 31.59,
          "usd_per_hour": 0.4962,
          "usd_per_million": 4.3623,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.9496
            },
            {
              "resource": "Memory",
              "quantity": "3.9 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 3.238,
              "usd_per_million": 0.1025
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 3.1652
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.1451
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "W53EpXULaAA38Bq6UyKAJfIDNnsnH9CaKD1H+cTQGR8/uJoj8c4p0NzrS2qXnJO288dCbXK4DEIp6IijjrL7DQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "FY1TSt1LPBJBlwdEvfp/pRjW+fV1nlE5hUljaYLmrRI="
        },
        "build_page": "/inferences/airavata/7b/mradermacher-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/mradermacher/airavata-gguf",
        "weeks": [
          "2026-W43",
          "2026-W45"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": 31.59,
        "context_tokens": 4096,
        "build": "Q4_K_M (4-bit) by mradermacher",
        "ref": "hf.co/mradermacher/Airavata-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 4.3623,
        "cost_breakdown": {
          "tokens_per_sec": 31.59,
          "usd_per_hour": 0.4962,
          "usd_per_million": 4.3623,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.9496
            },
            {
              "resource": "Memory",
              "quantity": "3.9 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 3.238,
              "usd_per_million": 0.1025
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 3.1652
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.1451
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "W53EpXULaAA38Bq6UyKAJfIDNnsnH9CaKD1H+cTQGR8/uJoj8c4p0NzrS2qXnJO288dCbXK4DEIp6IijjrL7DQ==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MzEuNTkzODQwNDMyMDg2MjIsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzODU5NTI5NTg4ODA5Mzc5LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwNGRiOWI3NmU3MjhjZTA1YTM1ZDZlYzdmNWI1ZjZkZmE2YTkzNzMyZmFkNDM2YjQ0MzYwNmVjNWMzMGE1YmU0NiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbnZpZGlhLzMyZyIsIm1vZGVsIjoiaGYuY28vbXJhZGVybWFjaGVyL0FpcmF2YXRhLUdHVUY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjEyOSwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjo2NDQuNzQ1OTI4OTk3OTk5OCwidG90YWxfdG9rZW5zIjoyMDM3MH19",
          "payload_sha256": "FY1TSt1LPBJBlwdEvfp/pRjW+fV1nlE5hUljaYLmrRI="
        },
        "build_page": "/inferences/airavata/7b/mradermacher-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/mradermacher/airavata-gguf",
        "weeks": [
          "2026-W43",
          "2026-W45"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "machine_kinds": 2,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W43",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 31.59,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by mradermacher",
          "ref": "hf.co/mradermacher/Airavata-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 4.3623,
          "cost_breakdown": {
            "tokens_per_sec": 31.59,
            "usd_per_hour": 0.4962,
            "usd_per_million": 4.3623,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.9496
              },
              {
                "resource": "Memory",
                "quantity": "3.9 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.238,
                "usd_per_million": 0.1025
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 3.1652
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.1451
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "W53EpXULaAA38Bq6UyKAJfIDNnsnH9CaKD1H+cTQGR8/uJoj8c4p0NzrS2qXnJO288dCbXK4DEIp6IijjrL7DQ==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MzEuNTkzODQwNDMyMDg2MjIsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzODU5NTI5NTg4ODA5Mzc5LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwNGRiOWI3NmU3MjhjZTA1YTM1ZDZlYzdmNWI1ZjZkZmE2YTkzNzMyZmFkNDM2YjQ0MzYwNmVjNWMzMGE1YmU0NiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbnZpZGlhLzMyZyIsIm1vZGVsIjoiaGYuY28vbXJhZGVybWFjaGVyL0FpcmF2YXRhLUdHVUY6UTRfS19NIiwibm90ZSI6InNpZ25lZCByb2xsaW5nIG9ic2VydmF0aW9uIGZyb20gcmVhbCBsb2NhbCBpbmZlcmVuY2UgbWV0cmljczsgQ0xJIGJlbmNoIGluZmVyZW5jZSByZW1haW5zIFNBLUJFTkNILTUiLCJydW50aW1lIjoib2xsYW1hLWhvc3QiLCJzYW1wbGVfY291bnQiOjEyOSwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjo2NDQuNzQ1OTI4OTk3OTk5OCwidG90YWxfdG9rZW5zIjoyMDM3MH19",
            "payload_sha256": "FY1TSt1LPBJBlwdEvfp/pRjW+fV1nlE5hUljaYLmrRI="
          },
          "build_page": "/inferences/airavata/7b/mradermacher-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/mradermacher/airavata-gguf",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": 4096,
          "build": "Q4_K_M (4-bit) by mradermacher",
          "ref": "hf.co/mradermacher/Airavata-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T21:33:27.602Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/airavata/7b/mradermacher-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/mradermacher/airavata-gguf",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Llama 2 community licence",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/ai4bharat/Airavata"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "32 tokens a second on Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Forge-1-Mini",
      "made_by": "north-ml1",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68955d7785865264dfdfa451/vEvWH_AFrN3g7Fg5H62zQ.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Text & chat",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/North-ML1/Forge-1-Mini-GGUF",
      "download_gb": null,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "reasons about code"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 1074.92,
        "context_tokens": null,
        "build": null,
        "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q2_K",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.0283,
        "cost_breakdown": {
          "tokens_per_sec": 1074.92,
          "usd_per_hour": 0.1095,
          "usd_per_million": 0.0283,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.0279
            },
            {
              "resource": "Memory",
              "quantity": "0.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0,
              "usd_per_million": 0
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0004
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T06:35:24.096Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "IrkEGr17iVZaj/ODrXupMsMKUHTnFJ1OBbe3wkcu745GdsFBTtKF/ay4Y7zgxgUX1dNyj9V9F5gyT4rXR8FuDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "CukUdyh0rDsTNqukQVnwOjBK7dhlbqriuHCqE73ThiI="
        },
        "build_page": "/inferences/north-1/forge-1-mini~north-ml1/north-ml1-q2_k-ollama",
        "hub_url": "https://huggingface.co/north-ml1/forge-1-mini-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 1074.92,
        "context_tokens": null,
        "build": null,
        "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q2_K",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.0283,
        "cost_breakdown": {
          "tokens_per_sec": 1074.92,
          "usd_per_hour": 0.1095,
          "usd_per_million": 0.0283,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.0279
            },
            {
              "resource": "Memory",
              "quantity": "0.0 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0,
              "usd_per_million": 0
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0004
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T06:35:24.096Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "IrkEGr17iVZaj/ODrXupMsMKUHTnFJ1OBbe3wkcu745GdsFBTtKF/ay4Y7zgxgUX1dNyj9V9F5gyT4rXR8FuDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "CukUdyh0rDsTNqukQVnwOjBK7dhlbqriuHCqE73ThiI="
        },
        "build_page": "/inferences/north-1/forge-1-mini~north-ml1/north-ml1-q2_k-ollama",
        "hub_url": "https://huggingface.co/north-ml1/forge-1-mini-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W43",
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 1074.92,
          "context_tokens": null,
          "build": null,
          "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q2_K",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.0283,
          "cost_breakdown": {
            "tokens_per_sec": 1074.92,
            "usd_per_hour": 0.1095,
            "usd_per_million": 0.0283,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.0279
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0004
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T06:35:24.096Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "IrkEGr17iVZaj/ODrXupMsMKUHTnFJ1OBbe3wkcu745GdsFBTtKF/ay4Y7zgxgUX1dNyj9V9F5gyT4rXR8FuDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "CukUdyh0rDsTNqukQVnwOjBK7dhlbqriuHCqE73ThiI="
          },
          "build_page": "/inferences/north-1/forge-1-mini~north-ml1/north-ml1-q2_k-ollama",
          "hub_url": "https://huggingface.co/north-ml1/forge-1-mini-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 628.82,
          "context_tokens": null,
          "build": null,
          "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q3_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.0484,
          "cost_breakdown": {
            "tokens_per_sec": 628.82,
            "usd_per_hour": 0.1095,
            "usd_per_million": 0.0484,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.0477
              },
              {
                "resource": "Memory",
                "quantity": "0.0 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0,
                "usd_per_million": 0
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0007
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "y/9+gykKIPCSJsYSEpvk9fIJQnUY75Um7tOW/5UIHGkPdZkAh2zu/zo427J6FaA4rDBIJaC/lY5GI3LcftmICQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "9j8M6DoaTc5FGJkESYl9qArmKrzf2bhtRyay8YPHL34="
          },
          "build_page": "/inferences/north-1/forge-1-mini~north-ml1/north-ml1-q3_k_m-ollama",
          "hub_url": "https://huggingface.co/north-ml1/forge-1-mini-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/North-ML1/Forge-1-Mini-GGUF"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "1075 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Granite 4.0 Speech-1B",
      "made_by": "ibm-granite",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/639bcaa2445b133a4e942436/CEW-OjXkRkDNmTxSu8Egh.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/granite-4-0/speech-1b",
      "hub_url": "https://huggingface.co/ibm-granite/granite-4.0-1b-speech",
      "download_gb": 1.1,
      "params_b": 1,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "returns valid JSON"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "Tesla T4",
          "gpu_count": 2,
          "vram_gb": 15,
          "host": "Kaggle",
          "vendor_org": "nvidia",
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
        "machines": 1,
        "tokens_per_sec": 129.7,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by ibm-granite",
        "ref": "hf.co/ibm-granite/granite-4.0-1b-speech-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 1.0444,
        "cost_breakdown": {
          "tokens_per_sec": 129.7,
          "usd_per_hour": 0.4877,
          "usd_per_million": 1.0444,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.2313
            },
            {
              "resource": "Memory",
              "quantity": "1.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.884,
              "usd_per_million": 0.0068
            },
            {
              "resource": "GPU",
              "quantity": "1 of 2 cards (the one this build fits on)",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.771
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.0353
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 312.8,
            "multi_core_hashes_per_sec": 618.9,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-02T21:33:01.078Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 2.4,
            "latency_ns": 21.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-02T21:33:01.673Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 21.6,
            "watts": 59.1,
            "tflops_per_watt": 0.365,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-02T21:43:16.805Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4293.1,
            "write_mb_s": 60.4,
            "random_read_iops": 514568,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-02T21:33:02.973Z",
            "measured_in": "TW_110",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-03T00:21:51.818Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220d79e72fae9f23f0437b58a3411e6bdb14e5c77f568cc3bdf6622534463c76d30",
          "signature": "Fm1ATdm6uyWZLgWqRl0N+H8lLhBqiphWUjfNmJc4WMOOjLCX5u/Npv3riQbav1c527iIF0XBBy9zxuIBWvcbDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "67asdisH59BWTjznW2/Hb+qCNI3+1+RDD8WWtarvqsI="
        },
        "build_page": "/inferences/granite-4-0/speech-1b/ibm-granite-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/ibm-granite/granite-4.0-1b-speech-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "c76d30"
        ]
      },
      "cheapest": {
        "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "Tesla T4",
          "gpu_count": 2,
          "vram_gb": 15,
          "host": "Kaggle",
          "vendor_org": "nvidia",
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
        "machines": 1,
        "tokens_per_sec": 129.7,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by ibm-granite",
        "ref": "hf.co/ibm-granite/granite-4.0-1b-speech-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 1.0444,
        "cost_breakdown": {
          "tokens_per_sec": 129.7,
          "usd_per_hour": 0.4877,
          "usd_per_million": 1.0444,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.2313
            },
            {
              "resource": "Memory",
              "quantity": "1.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.884,
              "usd_per_million": 0.0068
            },
            {
              "resource": "GPU",
              "quantity": "1 of 2 cards (the one this build fits on)",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.771
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.0353
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 312.8,
            "multi_core_hashes_per_sec": 618.9,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-02T21:33:01.078Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 2.4,
            "latency_ns": 21.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-02T21:33:01.673Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 21.6,
            "watts": 59.1,
            "tflops_per_watt": 0.365,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-02T21:43:16.805Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4293.1,
            "write_mb_s": 60.4,
            "random_read_iops": 514568,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-02T21:33:02.973Z",
            "measured_in": "TW_110",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-03T00:21:51.818Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220d79e72fae9f23f0437b58a3411e6bdb14e5c77f568cc3bdf6622534463c76d30",
          "signature": "Fm1ATdm6uyWZLgWqRl0N+H8lLhBqiphWUjfNmJc4WMOOjLCX5u/Npv3riQbav1c527iIF0XBBy9zxuIBWvcbDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "67asdisH59BWTjznW2/Hb+qCNI3+1+RDD8WWtarvqsI="
        },
        "build_page": "/inferences/granite-4-0/speech-1b/ibm-granite-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/ibm-granite/granite-4.0-1b-speech-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "c76d30"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "Tesla T4",
            "gpu_count": 2,
            "vram_gb": 15,
            "host": "Kaggle",
            "vendor_org": "nvidia",
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|Tesla T4|2|Kaggle",
          "machines": 1,
          "tokens_per_sec": 129.7,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by ibm-granite",
          "ref": "hf.co/ibm-granite/granite-4.0-1b-speech-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.0444,
          "cost_breakdown": {
            "tokens_per_sec": 129.7,
            "usd_per_hour": 0.4877,
            "usd_per_million": 1.0444,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.2313
              },
              {
                "resource": "Memory",
                "quantity": "1.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.884,
                "usd_per_million": 0.0068
              },
              {
                "resource": "GPU",
                "quantity": "1 of 2 cards (the one this build fits on)",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.771
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.0353
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 312.8,
              "multi_core_hashes_per_sec": 618.9,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T21:33:01.078Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 2.4,
              "latency_ns": 21.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T21:33:01.673Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 21.6,
              "watts": 59.1,
              "tflops_per_watt": 0.365,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-02T21:43:16.805Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4293.1,
              "write_mb_s": 60.4,
              "random_read_iops": 514568,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-02T21:33:02.973Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T00:21:51.818Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220d79e72fae9f23f0437b58a3411e6bdb14e5c77f568cc3bdf6622534463c76d30",
            "signature": "Fm1ATdm6uyWZLgWqRl0N+H8lLhBqiphWUjfNmJc4WMOOjLCX5u/Npv3riQbav1c527iIF0XBBy9zxuIBWvcbDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "67asdisH59BWTjznW2/Hb+qCNI3+1+RDD8WWtarvqsI="
          },
          "build_page": "/inferences/granite-4-0/speech-1b/ibm-granite-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/ibm-granite/granite-4.0-1b-speech-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "c76d30"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/ibm-granite/granite-4.0-1b-speech"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "130 tokens a second on Linux · 2× Tesla T4 15 GB · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "OpenHathi 7B Hindi base",
      "made_by": "sarvamai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/60270a7c32856987162c641a/umd13GCWVijwTDGZzw3q-.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/openhathi/7b",
      "hub_url": "https://huggingface.co/sarvamai/OpenHathi-7B-Hi-v0.1-Base",
      "download_gb": 3.9,
      "params_b": 7,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "reasons about code"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": 29.84,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by QuantFactory",
        "ref": "hf.co/QuantFactory/OpenHathi-7B-Hi-v0.1-Base-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 4.6183,
        "cost_breakdown": {
          "tokens_per_sec": 29.84,
          "usd_per_hour": 0.4962,
          "usd_per_million": 4.6183,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 1.0053
            },
            {
              "resource": "Memory",
              "quantity": "3.9 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 3.238,
              "usd_per_million": 0.1085
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 3.3509
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.1536
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "4IlGtATg8l/Qb5dr67qSNDegWE0907eRvB88tqfHR6Wlye9DQxogdwZUhStBF3KzthGy+viAubOFm+MKV7ySBA==",
          "signing_payload_b64": "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",
          "payload_sha256": "i+c8QRQgYxf87SX+zxUUe0MvxKgquPAGO0Wm6FAseSE="
        },
        "build_page": "/inferences/openhathi/7b/quantfactory-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/quantfactory/openhathi-7b-hi-v0.1-base-gguf",
        "weeks": [
          "2026-W43",
          "2026-W45"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": 29.84,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by QuantFactory",
        "ref": "hf.co/QuantFactory/OpenHathi-7B-Hi-v0.1-Base-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 4.6183,
        "cost_breakdown": {
          "tokens_per_sec": 29.84,
          "usd_per_hour": 0.4962,
          "usd_per_million": 4.6183,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 1.0053
            },
            {
              "resource": "Memory",
              "quantity": "3.9 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 3.238,
              "usd_per_million": 0.1085
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 3.3509
            },
            {
              "resource": "Energy",
              "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 4.583,
              "usd_per_million": 0.1536
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-05T06:18:49.588Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
          "signature": "4IlGtATg8l/Qb5dr67qSNDegWE0907eRvB88tqfHR6Wlye9DQxogdwZUhStBF3KzthGy+viAubOFm+MKV7ySBA==",
          "signing_payload_b64": "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",
          "payload_sha256": "i+c8QRQgYxf87SX+zxUUe0MvxKgquPAGO0Wm6FAseSE="
        },
        "build_page": "/inferences/openhathi/7b/quantfactory-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/quantfactory/openhathi-7b-hi-v0.1-base-gguf",
        "weeks": [
          "2026-W43",
          "2026-W45"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "machine_kinds": 2,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W43",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": 29.84,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by QuantFactory",
          "ref": "hf.co/QuantFactory/OpenHathi-7B-Hi-v0.1-Base-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 4.6183,
          "cost_breakdown": {
            "tokens_per_sec": 29.84,
            "usd_per_hour": 0.4962,
            "usd_per_million": 4.6183,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.0053
              },
              {
                "resource": "Memory",
                "quantity": "3.9 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 3.238,
                "usd_per_million": 0.1085
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 3.3509
              },
              {
                "resource": "Energy",
                "quantity": "165 W modelled (3 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 4.583,
                "usd_per_million": 0.1536
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T06:18:49.588Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112204db9b76e728ce05a35d6ec7f5b5f6dfa6a93732fad436b443606ec5c30a5be46",
            "signature": "4IlGtATg8l/Qb5dr67qSNDegWE0907eRvB88tqfHR6Wlye9DQxogdwZUhStBF3KzthGy+viAubOFm+MKV7ySBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "i+c8QRQgYxf87SX+zxUUe0MvxKgquPAGO0Wm6FAseSE="
          },
          "build_page": "/inferences/openhathi/7b/quantfactory-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/quantfactory/openhathi-7b-hi-v0.1-base-gguf",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by QuantFactory",
          "ref": "hf.co/QuantFactory/OpenHathi-7B-Hi-v0.1-Base-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T21:52:52.959Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/openhathi/7b/quantfactory-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/quantfactory/openhathi-7b-hi-v0.1-base-gguf",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Llama 2 community licence",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/sarvamai/OpenHathi-7B-Hi-v0.1-Base"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "30 tokens a second on Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "base",
      "made_by": "stabilityai",
      "maker_logo": "https://huggingface.co/avatars/24296b69e3199d9e76ae9fe684dd5c76.svg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Image generation",
      "community_build": false,
      "page": "/inferences/sd-turbo/base",
      "hub_url": "https://huggingface.co/Darylgeero/sd-turbo",
      "download_gb": 12.1,
      "params_b": 6.5,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "makes images"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP16 (16-bit, full precision) by Darylgeero",
        "ref": "Darylgeero/sd-turbo",
        "engine": "generation",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T17:52:12.506Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/sd-turbo/base/darylgeero-fp16-diffusers",
        "hub_url": "https://huggingface.co/darylgeero/sd-turbo",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by Darylgeero",
          "ref": "Darylgeero/sd-turbo",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T17:52:12.506Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sd-turbo/base/darylgeero-fp16-diffusers",
          "hub_url": "https://huggingface.co/darylgeero/sd-turbo",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": "https://huggingface.co/Darylgeero/sd-turbo"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "EmbeddingGemma 2 Base",
      "made_by": "google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Embeddings (search)",
      "community_build": false,
      "page": "/inferences/embeddinggemma-2/base",
      "hub_url": "https://huggingface.co/google/embeddinggemma-2",
      "download_gb": 0.2,
      "params_b": 0.3,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "search embeddings"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by ngquocvinh",
        "ref": "hf.co/ngquocvinh/EmbeddingGemma-2-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-06T12:40:34.040Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/embeddinggemma-2/base/ngquocvinh-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/ngquocvinh/embeddinggemma-2-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by ngquocvinh",
          "ref": "hf.co/ngquocvinh/EmbeddingGemma-2-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T12:40:34.040Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma-2/base/ngquocvinh-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/ngquocvinh/embeddinggemma-2-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/google/embeddinggemma-2"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "EmbeddingGemma 300M",
      "made_by": "google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Embeddings (search)",
      "community_build": false,
      "page": "/inferences/embeddinggemma/300m",
      "hub_url": "https://huggingface.co/google/embeddinggemma-300m",
      "download_gb": 0.2,
      "params_b": 0.3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "search embeddings"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "Q6_K (6-bit) by cduk",
        "ref": "hf.co/cduk/embeddinggemma-300m-GGUF-with-dense-modules:Q6_K",
        "engine": "ollama-host",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-25T20:35:41.959Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/embeddinggemma/300m~cduk-embeddinggemma-with-dense-modules/cduk-q6_k-ollama",
        "hub_url": "https://huggingface.co/cduk/embeddinggemma-300m-gguf-with-dense-modules",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": null,
      "machine_kinds": 7,
      "machines_total": 22,
      "weeks_measured": [
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q6_K (6-bit) by cduk",
          "ref": "hf.co/cduk/embeddinggemma-300m-GGUF-with-dense-modules:Q6_K",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T20:35:41.959Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m~cduk-embeddinggemma-with-dense-modules/cduk-q6_k-ollama",
          "hub_url": "https://huggingface.co/cduk/embeddinggemma-300m-gguf-with-dense-modules",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q8_0 (8-bit) by cduk",
          "ref": "hf.co/cduk/embeddinggemma-300m-GGUF-with-dense-modules:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T20:36:38.141Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m~cduk-embeddinggemma-with-dense-modules/cduk-q8_0-ollama",
          "hub_url": "https://huggingface.co/cduk/embeddinggemma-300m-gguf-with-dense-modules",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q2_K (2-bit) by gaianet",
          "ref": "hf.co/gaianet/embeddinggemma-300m-GGUF:Q2_K",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T20:31:59.132Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/gaianet-q2_k-ollama",
          "hub_url": "https://huggingface.co/gaianet/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q3_K_S (3-bit) by gaianet",
          "ref": "hf.co/gaianet/embeddinggemma-300m-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T20:32:24.192Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/gaianet-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/gaianet/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by gaianet",
          "ref": "hf.co/gaianet/embeddinggemma-300m-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T19:17:13.006Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/gaianet-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/gaianet/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by gaianet",
          "ref": "hf.co/gaianet/embeddinggemma-300m-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-26T04:18:19.570Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/gaianet-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/gaianet/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.22631",
            "cpus": 8,
            "cpu_model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
            "ram_gb": 7,
            "gpu_vendor": "AMD",
            "gpu_model": "AMD Radeon(TM) Vega 8 Mobile Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "amd",
            "host_org": null,
            "key": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by gaianet",
          "ref": "hf.co/gaianet/embeddinggemma-300m-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 997.6,
              "multi_core_hashes_per_sec": 8516.9,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T08:14:32.559Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "memory": {
              "total_gb": 7,
              "bandwidth_gb_s": 6,
              "latency_ns": 61.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T08:14:33.251Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2347.3,
              "write_mb_s": 204.2,
              "random_read_iops": 96070,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-10-29T08:14:33.869Z",
              "measured_in": "IN_560001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-28T13:19:30.256Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/gaianet-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/gaianet/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43",
            "2026-W44"
          ],
          "providers": [
            "d061c9"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q8_0 (8-bit) by ggml-org",
          "ref": "hf.co/ggml-org/embeddinggemma-300M-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T20:33:16.812Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/ggml-org-q8_0-ollama",
          "hub_url": "https://huggingface.co/ggml-org/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "F16 (16-bit, full precision) by lm-kit",
          "ref": "hf.co/lm-kit/embeddinggemma-300m-gguf:F16",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T20:37:13.045Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m~lm-kit-embeddinggemma/lm-kit-f16-ollama",
          "hub_url": "https://huggingface.co/lm-kit/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by lm-kit",
          "ref": "hf.co/lm-kit/embeddinggemma-300m-gguf:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T20:34:26.819Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m~lm-kit-embeddinggemma/lm-kit-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/lm-kit/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q5_K_M (5-bit) by lm-kit",
          "ref": "hf.co/lm-kit/embeddinggemma-300m-gguf:Q5_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T20:35:14.487Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m~lm-kit-embeddinggemma/lm-kit-q5_k_m-ollama",
          "hub_url": "https://huggingface.co/lm-kit/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q8_0 (8-bit) by lm-kit",
          "ref": "hf.co/lm-kit/embeddinggemma-300m-gguf:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T20:36:09.280Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m~lm-kit-embeddinggemma/lm-kit-q8_0-ollama",
          "hub_url": "https://huggingface.co/lm-kit/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q2_K (2-bit) by second-state",
          "ref": "hf.co/second-state/embeddinggemma-300m-GGUF:Q2_K",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T04:18:38.644Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/second-state-q2_k-ollama",
          "hub_url": "https://huggingface.co/second-state/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q3_K_S (3-bit) by second-state",
          "ref": "hf.co/second-state/embeddinggemma-300m-GGUF:Q3_K_S",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T04:18:35.022Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/second-state-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/second-state/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by second-state",
          "ref": "hf.co/second-state/embeddinggemma-300m-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T04:08:04.060Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/second-state-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/second-state/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by second-state",
          "ref": "hf.co/second-state/embeddinggemma-300m-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T12:39:41.676Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/second-state-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/second-state/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by second-state",
          "ref": "hf.co/second-state/embeddinggemma-300m-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T09:57:02.153Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/second-state-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/second-state/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by second-state",
          "ref": "hf.co/second-state/embeddinggemma-300m-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T04:25:34.835Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/second-state-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/second-state/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "BF16 (16-bit, full precision) by unsloth",
          "ref": "hf.co/unsloth/embeddinggemma-300m-GGUF:BF16",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T20:33:53.298Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/unsloth-bf16-ollama",
          "hub_url": "https://huggingface.co/unsloth/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "F32 (32-bit, full precision) by unsloth",
          "ref": "hf.co/unsloth/embeddinggemma-300m-GGUF:F32",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T23:47:58.918Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/unsloth-f32-ollama",
          "hub_url": "https://huggingface.co/unsloth/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · RTX PRO 6000 Blackwell Server Edition 96 GB · 48 CPUs · 177 GB RAM",
          "machine_detail": {
            "name": "Linux · RTX PRO 6000 Blackwell Server Edition 96 GB · 48 CPUs · 177 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 48,
            "cpu_model": "AMD EPYC 9B45",
            "ram_gb": 177,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "RTX PRO 6000 Blackwell Server Edition",
            "gpu_count": 1,
            "vram_gb": 96,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|48|177|nvidia|96|RTX PRO 6000 Blackwell Server Edition|1|Google Colab"
          },
          "machine_class": "linux|amd64|48|177|nvidia|96|RTX PRO 6000 Blackwell Server Edition|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "F32 (32-bit, full precision) by unsloth",
          "ref": "hf.co/unsloth/embeddinggemma-300m-GGUF:F32",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 9B45",
              "model_raw": "AMD EPYC 9B45",
              "clock_ghz": null,
              "cores_logical": 48,
              "cores_physical": 24,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 32,
              "single_core_hashes_per_sec": 1937.9,
              "multi_core_hashes_per_sec": 70559.4,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-03T02:23:41.449Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 176.9,
              "bandwidth_gb_s": 26.2,
              "latency_ns": 12,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-03T02:23:41.985Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA RTX PRO 6000 Blackwell Server Edition",
              "memory_gb": 95,
              "fp16_tflops": 372.6,
              "watts": 39.8,
              "tflops_per_watt": 9.353,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-03T02:33:49.172Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 15977,
              "write_mb_s": 36.1,
              "random_read_iops": 1825761,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-03T02:23:43.974Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T02:37:19.884Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/unsloth-f32-ollama",
          "hub_url": "https://huggingface.co/unsloth/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "e189d2"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q4_0 (4-bit) by unsloth",
          "ref": "hf.co/unsloth/embeddinggemma-300m-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T20:32:49.448Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/embeddinggemma/300m/unsloth-q4_0-ollama",
          "hub_url": "https://huggingface.co/unsloth/embeddinggemma-300m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Gemma terms of use",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/google/embeddinggemma-300m"
      },
      "builds": 16,
      "variants": [],
      "bottom_line": "Runs on 7 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "F5-Hindi 24 kHz (Spring Lab)",
      "made_by": "SWivid",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/636916bb85fef3ca96e2ef73/waHubUG9Ck-EBID9PBnWZ.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/f5-tts/hindi-24khz",
      "hub_url": "https://huggingface.co/SWivid/F5-TTS",
      "download_gb": 0.6,
      "params_b": 0.15,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 2,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by SPRINGLab",
        "ref": "SPRINGLab/F5-Hindi-24KHz",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:21:30.056Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/f5-tts/hindi-24khz/springlab-fp32-speech-hf",
        "hub_url": "https://huggingface.co/springlab/f5-hindi-24khz",
        "weeks": [
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "e3654f",
          "e61d3b"
        ]
      },
      "cheapest": null,
      "machine_kinds": 4,
      "machines_total": 5,
      "weeks_measured": [
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 2,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by SPRINGLab",
          "ref": "SPRINGLab/F5-Hindi-24KHz",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:21:30.056Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/f5-tts/hindi-24khz/springlab-fp32-speech-hf",
          "hub_url": "https://huggingface.co/springlab/f5-hindi-24khz",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e3654f",
            "e61d3b"
          ]
        },
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by SPRINGLab",
          "ref": "SPRINGLab/F5-Hindi-24KHz",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-28T22:07:07.724Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/f5-tts/hindi-24khz/springlab-fp32-speech-hf",
          "hub_url": "https://huggingface.co/springlab/f5-hindi-24khz",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by SPRINGLab",
          "ref": "SPRINGLab/F5-Hindi-24KHz",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T02:44:42.274Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/f5-tts/hindi-24khz/springlab-fp32-speech-hf",
          "hub_url": "https://huggingface.co/springlab/f5-hindi-24khz",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by SPRINGLab",
          "ref": "SPRINGLab/F5-Hindi-24KHz",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T18:57:18.035Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/f5-tts/hindi-24khz/springlab-fp32-speech-hf",
          "hub_url": "https://huggingface.co/springlab/f5-hindi-24khz",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/SWivid/F5-TTS"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 4 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "F5-TTS v1 Base (zero-shot voice, en/zh)",
      "made_by": "SWivid",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/636916bb85fef3ca96e2ef73/waHubUG9Ck-EBID9PBnWZ.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/f5-tts/v1-base",
      "hub_url": "https://huggingface.co/SWivid/F5-TTS",
      "download_gb": 6.3,
      "params_b": 0.34,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by SWivid",
        "ref": "SWivid/F5-TTS",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T02:44:12.556Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/f5-tts/v1-base/swivid-fp32-speech-hf",
        "hub_url": "https://huggingface.co/swivid/f5-tts",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by SWivid",
          "ref": "SWivid/F5-TTS",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T02:44:12.556Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/f5-tts/v1-base/swivid-fp32-speech-hf",
          "hub_url": "https://huggingface.co/swivid/f5-tts",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/SWivid/F5-TTS"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "kokoro base",
      "made_by": "hexgrad",
      "maker_logo": null,
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/kokoro/base",
      "hub_url": null,
      "download_gb": 0.3,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": "12th Gen Intel Core i3-12100",
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": "Intel(R) UHD Graphics 730",
          "gpu_count": 1,
          "vram_gb": 2,
          "host": "own machine",
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
        },
        "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "by kokoro",
        "ref": "kokoro",
        "engine": "speech-pod",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-04T00:33:59.425Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/kokoro/base/kokoro-unknown-speech-pod",
        "hub_url": null,
        "weeks": [
          "2026-W36",
          "2026-W38",
          "2026-W39",
          "2026-W43",
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 5,
      "machines_total": 6,
      "weeks_measured": [
        "2026-W36",
        "2026-W38",
        "2026-W39",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by kokoro",
          "ref": "kokoro",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-04T00:33:59.425Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/kokoro/base/kokoro-unknown-speech-pod",
          "hub_url": null,
          "weeks": [
            "2026-W36",
            "2026-W38",
            "2026-W39",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by kokoro",
          "ref": "kokoro",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-20T09:29:43.551Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/kokoro/base/kokoro-unknown-speech-pod",
          "hub_url": null,
          "weeks": [
            "2026-W36",
            "2026-W38",
            "2026-W39",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 2,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by kokoro",
          "ref": "kokoro",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-26T04:55:26.308Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/kokoro/base/kokoro-unknown-speech-pod",
          "hub_url": null,
          "weeks": [
            "2026-W36",
            "2026-W38",
            "2026-W39",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e61d3b",
            "e3654f"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by kokoro",
          "ref": "kokoro",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T21:18:23.009Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/kokoro/base/kokoro-unknown-speech-pod",
          "hub_url": null,
          "weeks": [
            "2026-W36",
            "2026-W38",
            "2026-W39",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by kokoro",
          "ref": "kokoro",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-12T18:25:57.138Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/kokoro/base/kokoro-unknown-speech-pod",
          "hub_url": null,
          "weeks": [
            "2026-W36",
            "2026-W38",
            "2026-W39",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 5 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "memorized-commons10",
      "made_by": "stabilityai",
      "maker_logo": "https://huggingface.co/avatars/f9387d0de88f54a189c061c52e0166ca.svg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Image generation",
      "community_build": false,
      "page": "/inferences/sd-turbo/memorized-commons10",
      "hub_url": "https://huggingface.co/nbkcnil/sd-turbo-memorized-commons10",
      "download_gb": 3.2,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "makes images"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "AMD EPYC 7B12",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "by nbkcnil",
        "ref": "nbkcnil/sd-turbo-memorized-commons10",
        "engine": "generation",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "AMD EPYC 7B12",
            "model_raw": "AMD EPYC 7B12",
            "clock_ghz": null,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 512,
            "l3_mb": 16,
            "single_core_hashes_per_sec": 1460.2,
            "multi_core_hashes_per_sec": 3246.1,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T01:38:20.212Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 6.3,
            "latency_ns": 21.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T01:38:21.135Z",
            "measured_in": "TW_110",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4982.5,
            "write_mb_s": 179.3,
            "random_read_iops": 476287,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-15T01:38:21.741Z",
            "measured_in": "TW_110",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-14T22:48:08.543Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/sd-turbo/memorized-commons10/nbkcnil-sd-turbo-memorized-commons10-diffusers",
        "hub_url": "https://huggingface.co/nbkcnil/sd-turbo-memorized-commons10",
        "weeks": [
          "2026-W46"
        ],
        "providers": [
          "662a4e"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "AMD EPYC 7B12",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by nbkcnil",
          "ref": "nbkcnil/sd-turbo-memorized-commons10",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T22:48:08.543Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sd-turbo/memorized-commons10/nbkcnil-sd-turbo-memorized-commons10-diffusers",
          "hub_url": "https://huggingface.co/nbkcnil/sd-turbo-memorized-commons10",
          "weeks": [
            "2026-W46"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/nbkcnil/sd-turbo-memorized-commons10"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MMS-TTS Assamese",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/mms-tts/as",
      "hub_url": "https://huggingface.co/facebook/mms-tts-asm",
      "download_gb": 0.1,
      "params_b": 0.04,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/mms-tts-asm",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:08:44.280Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/mms-tts/as/facebook-fp32-speech-hf",
        "hub_url": "https://huggingface.co/facebook/mms-tts-asm",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-asm",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:08:44.280Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/as/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-asm",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/mms-tts-asm"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MMS-TTS Bengali",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/mms-tts/bn",
      "hub_url": "https://huggingface.co/facebook/mms-tts-ben",
      "download_gb": 0.1,
      "params_b": 0.04,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/mms-tts-ben",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:02:45.086Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/mms-tts/bn/facebook-fp32-speech-hf",
        "hub_url": "https://huggingface.co/facebook/mms-tts-ben",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-ben",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:02:45.086Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/bn/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-ben",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/mms-tts-ben"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MMS-TTS Gujarati",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/mms-tts/gu",
      "hub_url": "https://huggingface.co/facebook/mms-tts-guj",
      "download_gb": 0.1,
      "params_b": 0.04,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/mms-tts-guj",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:05:19.858Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/mms-tts/gu/facebook-fp32-speech-hf",
        "hub_url": "https://huggingface.co/facebook/mms-tts-guj",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-guj",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:05:19.858Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/gu/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-guj",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/mms-tts-guj"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MMS-TTS Hindi",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/mms-tts/hi",
      "hub_url": "https://huggingface.co/facebook/mms-tts-hin",
      "download_gb": 0.1,
      "params_b": 0.04,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/mms-tts-hin",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:02:03.899Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/mms-tts/hi/facebook-fp32-speech-hf",
        "hub_url": "https://huggingface.co/facebook/mms-tts-hin",
        "weeks": [
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 5,
      "machines_total": 5,
      "weeks_measured": [
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-hin",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:02:03.899Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/hi/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-hin",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e3654f"
          ]
        },
        {
          "machine": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.22631",
            "cpus": 8,
            "cpu_model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
            "ram_gb": 7,
            "gpu_vendor": "AMD",
            "gpu_model": "AMD Radeon(TM) Vega 8 Mobile Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "amd",
            "host_org": null,
            "key": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-hin",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 997.6,
              "multi_core_hashes_per_sec": 8516.9,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T08:14:32.559Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "memory": {
              "total_gb": 7,
              "bandwidth_gb_s": 6,
              "latency_ns": 61.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T08:14:33.251Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2347.3,
              "write_mb_s": 204.2,
              "random_read_iops": 96070,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-10-29T08:14:33.869Z",
              "measured_in": "IN_560001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-28T12:02:55.721Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/hi/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-hin",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "d061c9"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-hin",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T09:29:06.742Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/hi/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-hin",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-hin",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T12:28:17.393Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/hi/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-hin",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-hin",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-08T15:09:20.684Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/hi/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-hin",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/mms-tts-hin"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 5 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MMS-TTS Kannada",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/mms-tts/kn",
      "hub_url": "https://huggingface.co/facebook/mms-tts-kan",
      "download_gb": 0.1,
      "params_b": 0.04,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/mms-tts-kan",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:05:55.617Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/mms-tts/kn/facebook-fp32-speech-hf",
        "hub_url": "https://huggingface.co/facebook/mms-tts-kan",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-kan",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:05:55.617Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/kn/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-kan",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/mms-tts-kan"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MMS-TTS Malayalam",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/mms-tts/ml",
      "hub_url": "https://huggingface.co/facebook/mms-tts-mal",
      "download_gb": 0.1,
      "params_b": 0.04,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/mms-tts-mal",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:06:35.047Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/mms-tts/ml/facebook-fp32-speech-hf",
        "hub_url": "https://huggingface.co/facebook/mms-tts-mal",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-mal",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:06:35.047Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/ml/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-mal",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/mms-tts-mal"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MMS-TTS Marathi",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/mms-tts/mr",
      "hub_url": "https://huggingface.co/facebook/mms-tts-mar",
      "download_gb": 0.1,
      "params_b": 0.04,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/mms-tts-mar",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:04:43.537Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/mms-tts/mr/facebook-fp32-speech-hf",
        "hub_url": "https://huggingface.co/facebook/mms-tts-mar",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-mar",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:04:43.537Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/mr/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-mar",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/mms-tts-mar"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MMS-TTS Odia",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/mms-tts/or",
      "hub_url": "https://huggingface.co/facebook/mms-tts-ory",
      "download_gb": 0.1,
      "params_b": 0.04,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/mms-tts-ory",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:08:07.736Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/mms-tts/or/facebook-fp32-speech-hf",
        "hub_url": "https://huggingface.co/facebook/mms-tts-ory",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-ory",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:08:07.736Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/or/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-ory",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/mms-tts-ory"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MMS-TTS Punjabi",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/mms-tts/pa",
      "hub_url": "https://huggingface.co/facebook/mms-tts-pan",
      "download_gb": 0.1,
      "params_b": 0.04,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/mms-tts-pan",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:07:16.475Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/mms-tts/pa/facebook-fp32-speech-hf",
        "hub_url": "https://huggingface.co/facebook/mms-tts-pan",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-pan",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:07:16.475Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/pa/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-pan",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/mms-tts-pan"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MMS-TTS Tamil",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/mms-tts/ta",
      "hub_url": "https://huggingface.co/facebook/mms-tts-tam",
      "download_gb": 0.1,
      "params_b": 0.04,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/mms-tts-tam",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:03:59.964Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/mms-tts/ta/facebook-fp32-speech-hf",
        "hub_url": "https://huggingface.co/facebook/mms-tts-tam",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-tam",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:03:59.964Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/ta/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-tam",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/mms-tts-tam"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MMS-TTS Telugu",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/mms-tts/te",
      "hub_url": "https://huggingface.co/facebook/mms-tts-tel",
      "download_gb": 0.1,
      "params_b": 0.04,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/mms-tts-tel",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:03:21.754Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/mms-tts/te/facebook-fp32-speech-hf",
        "hub_url": "https://huggingface.co/facebook/mms-tts-tel",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-tel",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:03:21.754Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/te/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-tel",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/mms-tts-tel"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MMS-TTS Urdu",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/mms-tts/ur",
      "hub_url": "https://huggingface.co/facebook/mms-tts-urd-script_arabic",
      "download_gb": 0.1,
      "params_b": 0.04,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/mms-tts-urd-script_arabic",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:09:39.060Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/mms-tts/ur/facebook-fp32-speech-hf",
        "hub_url": "https://huggingface.co/facebook/mms-tts-urd-script_arabic",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/mms-tts-urd-script_arabic",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:09:39.060Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/mms-tts/ur/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/mms-tts-urd-script_arabic",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/mms-tts-urd-script_arabic"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "moonshine base",
      "made_by": "UsefulSensors",
      "maker_logo": null,
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Speech to text",
      "community_build": false,
      "page": "/inferences/moonshine/base",
      "hub_url": "https://huggingface.co/UsefulSensors/moonshine-base",
      "download_gb": 0.2,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "transcribes"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "by UsefulSensors",
        "ref": "UsefulSensors/moonshine-base",
        "engine": "speech-pod",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-20T09:29:43.546Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/moonshine/base/usefulsensors-moonshine-base-speech-pod",
        "hub_url": "https://huggingface.co/usefulsensors/moonshine-base",
        "weeks": [
          "2026-W43",
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 4,
      "machines_total": 5,
      "weeks_measured": [
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by UsefulSensors",
          "ref": "UsefulSensors/moonshine-base",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-20T09:29:43.546Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/moonshine/base/usefulsensors-moonshine-base-speech-pod",
          "hub_url": "https://huggingface.co/usefulsensors/moonshine-base",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 2,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by UsefulSensors",
          "ref": "UsefulSensors/moonshine-base",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-26T04:55:26.281Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/moonshine/base/usefulsensors-moonshine-base-speech-pod",
          "hub_url": "https://huggingface.co/usefulsensors/moonshine-base",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e61d3b",
            "e3654f"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by UsefulSensors",
          "ref": "UsefulSensors/moonshine-base",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T22:16:56.943Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/moonshine/base/usefulsensors-moonshine-base-speech-pod",
          "hub_url": "https://huggingface.co/usefulsensors/moonshine-base",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by UsefulSensors",
          "ref": "UsefulSensors/moonshine-base",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T13:21:38.168Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/moonshine/base/usefulsensors-moonshine-base-speech-pod",
          "hub_url": "https://huggingface.co/usefulsensors/moonshine-base",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/UsefulSensors/moonshine-base"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 4 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "moonshine tiny",
      "made_by": "UsefulSensors",
      "maker_logo": null,
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Speech to text",
      "community_build": false,
      "page": "/inferences/moonshine/tiny",
      "hub_url": "https://huggingface.co/UsefulSensors/moonshine-tiny",
      "download_gb": 0.1,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "transcribes"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 2,
          "host": null,
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|2|||"
        },
        "machine_class": "windows|amd64|8|16|intel|2|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "by UsefulSensors",
        "ref": "UsefulSensors/moonshine-tiny",
        "engine": "speech-pod",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-09-21T08:26:20.102Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/moonshine/tiny/usefulsensors-moonshine-tiny-speech-pod",
        "hub_url": "https://huggingface.co/usefulsensors/moonshine-tiny",
        "weeks": [
          "2026-W30",
          "2026-W37",
          "2026-W39",
          "2026-W43",
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 5,
      "machines_total": 6,
      "weeks_measured": [
        "2026-W30",
        "2026-W37",
        "2026-W39",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by UsefulSensors",
          "ref": "UsefulSensors/moonshine-tiny",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-21T08:26:20.102Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/moonshine/tiny/usefulsensors-moonshine-tiny-speech-pod",
          "hub_url": "https://huggingface.co/usefulsensors/moonshine-tiny",
          "weeks": [
            "2026-W30",
            "2026-W37",
            "2026-W39",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by UsefulSensors",
          "ref": "UsefulSensors/moonshine-tiny",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-19T21:17:02.974Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/moonshine/tiny/usefulsensors-moonshine-tiny-speech-pod",
          "hub_url": "https://huggingface.co/usefulsensors/moonshine-tiny",
          "weeks": [
            "2026-W30",
            "2026-W37",
            "2026-W39",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 2,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by UsefulSensors",
          "ref": "UsefulSensors/moonshine-tiny",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-26T04:50:38.161Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/moonshine/tiny/usefulsensors-moonshine-tiny-speech-pod",
          "hub_url": "https://huggingface.co/usefulsensors/moonshine-tiny",
          "weeks": [
            "2026-W30",
            "2026-W37",
            "2026-W39",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e61d3b",
            "e3654f"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by UsefulSensors",
          "ref": "UsefulSensors/moonshine-tiny",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T22:17:23.023Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/moonshine/tiny/usefulsensors-moonshine-tiny-speech-pod",
          "hub_url": "https://huggingface.co/usefulsensors/moonshine-tiny",
          "weeks": [
            "2026-W30",
            "2026-W37",
            "2026-W39",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by UsefulSensors",
          "ref": "UsefulSensors/moonshine-tiny",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-12T18:32:28.736Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/moonshine/tiny/usefulsensors-moonshine-tiny-speech-pod",
          "hub_url": "https://huggingface.co/usefulsensors/moonshine-tiny",
          "weeks": [
            "2026-W30",
            "2026-W37",
            "2026-W39",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/UsefulSensors/moonshine-tiny"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 5 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MusicGen Medium",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Music & sound",
      "community_build": false,
      "page": "/inferences/musicgen/medium",
      "hub_url": "https://huggingface.co/facebook/musicgen-medium",
      "download_gb": 11.1,
      "params_b": 1.5,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "makes music"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/musicgen-medium",
        "engine": "audiogen",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-22T09:35:49.278Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/musicgen/medium/facebook-fp32-audiogen",
        "hub_url": "https://huggingface.co/facebook/musicgen-medium",
        "weeks": [
          "2026-W43",
          "2026-W45"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 2,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W43",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/musicgen-medium",
          "engine": "audiogen",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-22T09:35:49.278Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/musicgen/medium/facebook-fp32-audiogen",
          "hub_url": "https://huggingface.co/facebook/musicgen-medium",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/musicgen-medium",
          "engine": "audiogen",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T02:58:25.849Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/musicgen/medium/facebook-fp32-audiogen",
          "hub_url": "https://huggingface.co/facebook/musicgen-medium",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/musicgen-medium"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 2 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "MusicGen Small",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Music & sound",
      "community_build": false,
      "page": "/inferences/musicgen/small",
      "hub_url": "https://huggingface.co/facebook/musicgen-small",
      "download_gb": 2.2,
      "params_b": 0.3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "makes music"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 2,
          "host": null,
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|2|||"
        },
        "machine_class": "windows|amd64|8|16|intel|2|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/musicgen-small",
        "engine": "audiogen",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-09-25T03:32:40.776Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/musicgen/small/facebook-fp32-audiogen",
        "hub_url": "https://huggingface.co/facebook/musicgen-small",
        "weeks": [
          "2026-W36",
          "2026-W37",
          "2026-W38",
          "2026-W39",
          "2026-W40",
          "2026-W41",
          "2026-W43",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 11,
      "machines_total": 11,
      "weeks_measured": [
        "2026-W36",
        "2026-W37",
        "2026-W38",
        "2026-W39",
        "2026-W40",
        "2026-W41",
        "2026-W43",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/musicgen-small",
          "engine": "audiogen",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-25T03:32:40.776Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/musicgen/small/facebook-fp32-audiogen",
          "hub_url": "https://huggingface.co/facebook/musicgen-small",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/musicgen-small",
          "engine": "audiogen",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-13T23:07:32.401Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/musicgen/small/facebook-fp32-audiogen",
          "hub_url": "https://huggingface.co/facebook/musicgen-small",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/musicgen-small",
          "engine": "audiogen",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-10T21:33:47.641Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/musicgen/small/facebook-fp32-audiogen",
          "hub_url": "https://huggingface.co/facebook/musicgen-small",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Mac · no GPU · 8 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Mac · no GPU · 8 CPUs · 8 GB RAM",
            "os": "Mac",
            "os_version": "macOS 15.7.3",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|8|8|none|0|||"
          },
          "machine_class": "darwin|arm64|8|8|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/musicgen-small",
          "engine": "audiogen",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-06T21:46:31.373Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/musicgen/small/facebook-fp32-audiogen",
          "hub_url": "https://huggingface.co/facebook/musicgen-small",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "0c5265"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/musicgen-small",
          "engine": "audiogen",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-02T17:12:56.374Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/musicgen/small/facebook-fp32-audiogen",
          "hub_url": "https://huggingface.co/facebook/musicgen-small",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · no GPU · 8 CPUs · 8 GB RAM",
          "machine_detail": {
            "name": "Windows · no GPU · 8 CPUs · 8 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 8,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "windows|amd64|8|8|none|0|||"
          },
          "machine_class": "windows|amd64|8|8|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/musicgen-small",
          "engine": "audiogen",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-09T11:55:29.794Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/musicgen/small/facebook-fp32-audiogen",
          "hub_url": "https://huggingface.co/facebook/musicgen-small",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "12954d"
          ]
        },
        {
          "machine": "Windows · NVIDIA 4 GB · 12 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 4 GB · 12 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 12,
            "cpu_model": null,
            "ram_gb": 7,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 4,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|12|7|nvidia|4|||"
          },
          "machine_class": "windows|amd64|12|7|nvidia|4|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/musicgen-small",
          "engine": "audiogen",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-16T12:17:47.739Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/musicgen/small/facebook-fp32-audiogen",
          "hub_url": "https://huggingface.co/facebook/musicgen-small",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9a1db0"
          ]
        },
        {
          "machine": "Windows · AMD 1 GB · 8 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · AMD 1 GB · 8 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.22631",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 7,
            "gpu_vendor": "AMD",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 1,
            "host": null,
            "vendor_org": "amd",
            "host_org": null,
            "key": "windows|amd64|8|7|amd|1|||"
          },
          "machine_class": "windows|amd64|8|7|amd|1|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/musicgen-small",
          "engine": "audiogen",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 997.6,
              "multi_core_hashes_per_sec": 8516.9,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T08:14:32.559Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "memory": {
              "total_gb": 7,
              "bandwidth_gb_s": 6,
              "latency_ns": 61.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T08:14:33.251Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2347.3,
              "write_mb_s": 204.2,
              "random_read_iops": 96070,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-10-29T08:14:33.869Z",
              "measured_in": "IN_560001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-17T08:58:26.913Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/musicgen/small/facebook-fp32-audiogen",
          "hub_url": "https://huggingface.co/facebook/musicgen-small",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "d061c9"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/musicgen-small",
          "engine": "audiogen",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-22T07:59:28.315Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/musicgen/small/facebook-fp32-audiogen",
          "hub_url": "https://huggingface.co/facebook/musicgen-small",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/musicgen-small",
          "engine": "audiogen",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T02:55:08.605Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/musicgen/small/facebook-fp32-audiogen",
          "hub_url": "https://huggingface.co/facebook/musicgen-small",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "AMD EPYC 7B12",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/musicgen-small",
          "engine": "audiogen",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T22:39:24.329Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/musicgen/small/facebook-fp32-audiogen",
          "hub_url": "https://huggingface.co/facebook/musicgen-small",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W43",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/musicgen-small"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 11 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "piper-voices",
      "made_by": "rhasspy",
      "maker_logo": null,
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Other",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/rhasspy/piper-voices",
      "download_gb": null,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [
        "English"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": "12th Gen Intel Core i3-12100",
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": "Intel(R) UHD Graphics 730",
          "gpu_count": 1,
          "vram_gb": 2,
          "host": "own machine",
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
        },
        "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "rhasspy/piper-voices",
        "engine": "speech-pod",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-04T00:40:01.276Z",
        "signed": true,
        "proof": null,
        "build_page": null,
        "hub_url": "https://huggingface.co/rhasspy/piper-voices",
        "weeks": [
          "2026-W34",
          "2026-W37",
          "2026-W39",
          "2026-W44",
          "2026-W45"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W34",
        "2026-W37",
        "2026-W39",
        "2026-W44",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "rhasspy/piper-voices",
          "engine": "speech-pod",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-04T00:40:01.276Z",
          "signed": true,
          "proof": null,
          "build_page": null,
          "hub_url": "https://huggingface.co/rhasspy/piper-voices",
          "weeks": [
            "2026-W34",
            "2026-W37",
            "2026-W39",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "9e3c79"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "SAM ViT-Base",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Vision (counting, locating)",
      "community_build": false,
      "page": "/inferences/sam/vit-base",
      "hub_url": "https://huggingface.co/facebook/sam-vit-base",
      "download_gb": 0.7,
      "params_b": 0.09,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "counts objects"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": "12th Gen Intel Core i3-12100",
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": "Intel(R) UHD Graphics 730",
          "gpu_count": 1,
          "vram_gb": 2,
          "host": "own machine",
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
        },
        "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/sam-vit-base",
        "engine": "vision-task",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-28T15:55:21.008Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/sam/vit-base/facebook-fp32-vision-task",
        "hub_url": "https://huggingface.co/facebook/sam-vit-base",
        "weeks": [
          "2026-W36",
          "2026-W37",
          "2026-W38",
          "2026-W40",
          "2026-W43",
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 9,
      "machines_total": 9,
      "weeks_measured": [
        "2026-W36",
        "2026-W37",
        "2026-W38",
        "2026-W40",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/sam-vit-base",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-28T15:55:21.008Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sam/vit-base/facebook-fp32-vision-task",
          "hub_url": "https://huggingface.co/facebook/sam-vit-base",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/sam-vit-base",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-14T05:53:34.018Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sam/vit-base/facebook-fp32-vision-task",
          "hub_url": "https://huggingface.co/facebook/sam-vit-base",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/sam-vit-base",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-01T08:42:58.547Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sam/vit-base/facebook-fp32-vision-task",
          "hub_url": "https://huggingface.co/facebook/sam-vit-base",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/sam-vit-base",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-22T08:16:04.272Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sam/vit-base/facebook-fp32-vision-task",
          "hub_url": "https://huggingface.co/facebook/sam-vit-base",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/sam-vit-base",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T11:39:43.587Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sam/vit-base/facebook-fp32-vision-task",
          "hub_url": "https://huggingface.co/facebook/sam-vit-base",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.22631",
            "cpus": 8,
            "cpu_model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
            "ram_gb": 7,
            "gpu_vendor": "AMD",
            "gpu_model": "AMD Radeon(TM) Vega 8 Mobile Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "amd",
            "host_org": null,
            "key": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/sam-vit-base",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 997.6,
              "multi_core_hashes_per_sec": 8516.9,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T08:14:32.559Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "memory": {
              "total_gb": 7,
              "bandwidth_gb_s": 6,
              "latency_ns": 61.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T08:14:33.251Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2347.3,
              "write_mb_s": 204.2,
              "random_read_iops": 96070,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-10-29T08:14:33.869Z",
              "measured_in": "IN_560001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-28T12:16:19.470Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sam/vit-base/facebook-fp32-vision-task",
          "hub_url": "https://huggingface.co/facebook/sam-vit-base",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "d061c9"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/sam-vit-base",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T17:10:46.263Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sam/vit-base/facebook-fp32-vision-task",
          "hub_url": "https://huggingface.co/facebook/sam-vit-base",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "AMD EPYC 7B12",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/sam-vit-base",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T21:29:30.817Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sam/vit-base/facebook-fp32-vision-task",
          "hub_url": "https://huggingface.co/facebook/sam-vit-base",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/sam-vit-base",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T13:30:31.227Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sam/vit-base/facebook-fp32-vision-task",
          "hub_url": "https://huggingface.co/facebook/sam-vit-base",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/facebook/sam-vit-base"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 9 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Sana 1.6B 1024px",
      "made_by": "Efficient-Large-Model",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62b4b5beb25cb80fcf278354/SIddx3hYu5rWXA4-O3Oaj.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Image generation",
      "community_build": false,
      "page": "/inferences/sana/1600m-1024px",
      "hub_url": "https://huggingface.co/Efficient-Large-Model/Sana_1600M_1024px_diffusers",
      "download_gb": 3,
      "params_b": 1.6,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "makes images"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 2,
          "host": null,
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|2|||"
        },
        "machine_class": "windows|amd64|8|16|intel|2|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP16 (16-bit, full precision) by Efficient-Large-Model",
        "ref": "Efficient-Large-Model/Sana_1600M_1024px_diffusers",
        "engine": "generation",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-09-13T18:37:04.807Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/sana/1600m-1024px/efficient-large-model-fp16-diffusers",
        "hub_url": "https://huggingface.co/efficient-large-model/sana_1600m_1024px_diffusers",
        "weeks": [
          "2026-W36",
          "2026-W37",
          "2026-W38",
          "2026-W43",
          "2026-W45"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 4,
      "machines_total": 4,
      "weeks_measured": [
        "2026-W36",
        "2026-W37",
        "2026-W38",
        "2026-W43",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by Efficient-Large-Model",
          "ref": "Efficient-Large-Model/Sana_1600M_1024px_diffusers",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-13T18:37:04.807Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sana/1600m-1024px/efficient-large-model-fp16-diffusers",
          "hub_url": "https://huggingface.co/efficient-large-model/sana_1600m_1024px_diffusers",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by Efficient-Large-Model",
          "ref": "Efficient-Large-Model/Sana_1600M_1024px_diffusers",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-14T04:29:46.125Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sana/1600m-1024px/efficient-large-model-fp16-diffusers",
          "hub_url": "https://huggingface.co/efficient-large-model/sana_1600m_1024px_diffusers",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by Efficient-Large-Model",
          "ref": "Efficient-Large-Model/Sana_1600M_1024px_diffusers",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-22T08:18:05.273Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sana/1600m-1024px/efficient-large-model-fp16-diffusers",
          "hub_url": "https://huggingface.co/efficient-large-model/sana_1600m_1024px_diffusers",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by Efficient-Large-Model",
          "ref": "Efficient-Large-Model/Sana_1600M_1024px_diffusers",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T22:22:32.932Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sana/1600m-1024px/efficient-large-model-fp16-diffusers",
          "hub_url": "https://huggingface.co/efficient-large-model/sana_1600m_1024px_diffusers",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Efficient-Large-Model/Sana_1600M_1024px_diffusers"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 4 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Sana_600M_1024px_diffusers",
      "made_by": "beh1222",
      "maker_logo": "https://huggingface.co/avatars/4b4e596e0d2561683a408713a8963ed4.svg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Other",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/Beh1222/Sana_600M_1024px_diffusers",
      "download_gb": null,
      "params_b": 0.6,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "makes images"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "Beh1222/Sana_600M_1024px_diffusers",
        "engine": "generation",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-03T22:23:26.066Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/sana/600m/beh1222-fp16-diffusers",
        "hub_url": "https://huggingface.co/beh1222/sana_600m_1024px_diffusers",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "Beh1222/Sana_600M_1024px_diffusers",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T22:23:26.066Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sana/600m/beh1222-fp16-diffusers",
          "hub_url": "https://huggingface.co/beh1222/sana_600m_1024px_diffusers",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Beh1222/Sana_600M_1024px_diffusers"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "SD-Turbo (SD 2.1 distilled)",
      "made_by": "stabilityai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/643feeb67bc3fbde1385cc25/7vmYr2XwVcPtkLzac_jxQ.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Image generation",
      "community_build": false,
      "page": "/inferences/sd-turbo/2-1",
      "hub_url": "https://huggingface.co/stabilityai/sd-turbo",
      "download_gb": 2.4,
      "params_b": 0.9,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "makes images"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 2,
          "host": null,
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|2|||"
        },
        "machine_class": "windows|amd64|8|16|intel|2|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP16 (16-bit, full precision) by stabilityai",
        "ref": "stabilityai/sd-turbo",
        "engine": "generation",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-09-15T21:07:30.512Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/sd-turbo/2-1/stabilityai-fp16-diffusers",
        "hub_url": "https://huggingface.co/stabilityai/sd-turbo",
        "weeks": [
          "2026-W37",
          "2026-W38",
          "2026-W43",
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 6,
      "machines_total": 7,
      "weeks_measured": [
        "2026-W37",
        "2026-W38",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by stabilityai",
          "ref": "stabilityai/sd-turbo",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-15T21:07:30.512Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sd-turbo/2-1/stabilityai-fp16-diffusers",
          "hub_url": "https://huggingface.co/stabilityai/sd-turbo",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by stabilityai",
          "ref": "stabilityai/sd-turbo",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-14T04:05:59.345Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sd-turbo/2-1/stabilityai-fp16-diffusers",
          "hub_url": "https://huggingface.co/stabilityai/sd-turbo",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by stabilityai",
          "ref": "stabilityai/sd-turbo",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-22T08:16:35.492Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sd-turbo/2-1/stabilityai-fp16-diffusers",
          "hub_url": "https://huggingface.co/stabilityai/sd-turbo",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by stabilityai",
          "ref": "stabilityai/sd-turbo",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T04:54:26.416Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sd-turbo/2-1/stabilityai-fp16-diffusers",
          "hub_url": "https://huggingface.co/stabilityai/sd-turbo",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by stabilityai",
          "ref": "stabilityai/sd-turbo",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T21:24:17.998Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sd-turbo/2-1/stabilityai-fp16-diffusers",
          "hub_url": "https://huggingface.co/stabilityai/sd-turbo",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 2,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by stabilityai",
          "ref": "stabilityai/sd-turbo",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-11T06:16:14.890Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sd-turbo/2-1/stabilityai-fp16-diffusers",
          "hub_url": "https://huggingface.co/stabilityai/sd-turbo",
          "weeks": [
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e",
            "4e94b2"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": "https://huggingface.co/stabilityai/sd-turbo"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 6 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "sd-turbo-phi-lattice",
      "made_by": "ununununununununununununununununununun",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6460e9ec9aad9e149722e9ef/DqDb6pLUzI8GOo5R1PqL7.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Other",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/ununununununununununununununununununun/sd-turbo-phi-lattice",
      "download_gb": null,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "makes images"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "ununununununununununununununununununun/sd-turbo-phi-lattice",
        "engine": "generation",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-06T09:21:26.590Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/sd-turbo/phi-lattice/ununununununununununununununununununun-sd-turbo-phi-lattice-diffusers",
        "hub_url": "https://huggingface.co/ununununununununununununununununununun/sd-turbo-phi-lattice",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": null,
      "machine_kinds": 3,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "ununununununununununununununununununun/sd-turbo-phi-lattice",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T09:21:26.590Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sd-turbo/phi-lattice/ununununununununununununununununununun-sd-turbo-phi-lattice-diffusers",
          "hub_url": "https://huggingface.co/ununununununununununununununununununun/sd-turbo-phi-lattice",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "ununununununununununununununununununun/sd-turbo-phi-lattice",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T20:41:58.319Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sd-turbo/phi-lattice/ununununununununununununununununununun-sd-turbo-phi-lattice-diffusers",
          "hub_url": "https://huggingface.co/ununununununununununununununununununun/sd-turbo-phi-lattice",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "ununununununununununununununununununun/sd-turbo-phi-lattice",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-08T01:14:59.288Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sd-turbo/phi-lattice/ununununununununununununununununununun-sd-turbo-phi-lattice-diffusers",
          "hub_url": "https://huggingface.co/ununununununununununununununununununun/sd-turbo-phi-lattice",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/ununununununununununununununununununun/sd-turbo-phi-lattice"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 3 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "sd-turbo-phi-lattice-v2",
      "made_by": "ununununununununununununununununununun",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6460e9ec9aad9e149722e9ef/DqDb6pLUzI8GOo5R1PqL7.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Other",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/ununununununununununununununununununun/sd-turbo-phi-lattice-v2",
      "download_gb": null,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "makes images"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "ununununununununununununununununununun/sd-turbo-phi-lattice-v2",
        "engine": "generation",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-06T09:24:30.962Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/sd-turbo/phi-lattice-v2/ununununununununununununununununununun-sd-turbo-phi-lattice-v2-diffusers",
        "hub_url": "https://huggingface.co/ununununununununununununununununununun/sd-turbo-phi-lattice-v2",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": null,
      "machine_kinds": 2,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "ununununununununununununununununununun/sd-turbo-phi-lattice-v2",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T09:24:30.962Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sd-turbo/phi-lattice-v2/ununununununununununununununununununun-sd-turbo-phi-lattice-v2-diffusers",
          "hub_url": "https://huggingface.co/ununununununununununununununununununun/sd-turbo-phi-lattice-v2",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "ununununununununununununununununununun/sd-turbo-phi-lattice-v2",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T22:22:41.059Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sd-turbo/phi-lattice-v2/ununununununununununununununununununun-sd-turbo-phi-lattice-v2-diffusers",
          "hub_url": "https://huggingface.co/ununununununununununununununununununun/sd-turbo-phi-lattice-v2",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "2e5c02"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/ununununununununununununununununununun/sd-turbo-phi-lattice-v2"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 2 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Seamless M4T v2 Large (speech → text)",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Speech to text",
      "community_build": false,
      "page": "/inferences/seamless-m4t/v2-large",
      "hub_url": "https://huggingface.co/facebook/seamless-m4t-v2-large",
      "download_gb": 8.6,
      "params_b": 2.3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "transcribes"
      ],
      "languages_passed": [
        "Assamese",
        "Bengali",
        "English",
        "Gujarati",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Nepali",
        "Odia",
        "Punjabi",
        "Tamil",
        "Telugu",
        "Urdu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/seamless-m4t-v2-large",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-06T18:06:00.315Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/seamless-m4t/v2-large/facebook-fp32-speech-hf",
        "hub_url": "https://huggingface.co/facebook/seamless-m4t-v2-large",
        "weeks": [
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": null,
      "machine_kinds": 3,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/seamless-m4t-v2-large",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T18:06:00.315Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/seamless-m4t/v2-large/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/seamless-m4t-v2-large",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/seamless-m4t-v2-large",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-08T18:44:18.195Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/seamless-m4t/v2-large/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/seamless-m4t-v2-large",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/seamless-m4t-v2-large",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T13:02:09.694Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/seamless-m4t/v2-large/facebook-fp32-speech-hf",
          "hub_url": "https://huggingface.co/facebook/seamless-m4t-v2-large",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/seamless-m4t-v2-large"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 3 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "spkrec-ecapa-voxceleb",
      "made_by": "speechbrain",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1663000279893-60243f18c1f3c79f98e4b382.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Speaker recognition",
      "community_build": false,
      "page": "/inferences/ecapa-tdnn/voxceleb",
      "hub_url": "https://huggingface.co/speechbrain/spkrec-ecapa-voxceleb",
      "download_gb": 0.1,
      "params_b": 0.02,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "identifies speakers"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by speechbrain",
        "ref": "speechbrain/spkrec-ecapa-voxceleb",
        "engine": "speaker-id",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T14:30:17.679Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/ecapa-tdnn/voxceleb/speechbrain-fp32-speaker-id",
        "hub_url": "https://huggingface.co/speechbrain/spkrec-ecapa-voxceleb",
        "weeks": [
          "2026-W43",
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 3,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by speechbrain",
          "ref": "speechbrain/spkrec-ecapa-voxceleb",
          "engine": "speaker-id",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T14:30:17.679Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/ecapa-tdnn/voxceleb/speechbrain-fp32-speaker-id",
          "hub_url": "https://huggingface.co/speechbrain/spkrec-ecapa-voxceleb",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e3654f"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by speechbrain",
          "ref": "speechbrain/spkrec-ecapa-voxceleb",
          "engine": "speaker-id",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T17:02:07.301Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/ecapa-tdnn/voxceleb/speechbrain-fp32-speaker-id",
          "hub_url": "https://huggingface.co/speechbrain/spkrec-ecapa-voxceleb",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "AMD EPYC 7B12",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by speechbrain",
          "ref": "speechbrain/spkrec-ecapa-voxceleb",
          "engine": "speaker-id",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T20:02:52.174Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/ecapa-tdnn/voxceleb/speechbrain-fp32-speaker-id",
          "hub_url": "https://huggingface.co/speechbrain/spkrec-ecapa-voxceleb",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/speechbrain/spkrec-ecapa-voxceleb"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 3 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "t2v-1.3b",
      "made_by": "Wan-AI",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/67b610677ea7952def8b29c6/N6jQbbeaa_FcUY-wI1dgG.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Video generation",
      "community_build": false,
      "page": "/inferences/wan-2/t2v-1.3b",
      "hub_url": "https://huggingface.co/Wan-AI/Wan2.1-T2V-1.3B-Diffusers",
      "download_gb": 26.9,
      "params_b": 1.3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "makes video"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "by Wan-AI",
        "ref": "Wan-AI/Wan2.1-T2V-1.3B-Diffusers",
        "engine": "diffusers",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-22T08:04:30.830Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/wan-2/t2v-1-3b/wan-ai-bf16-diffusers",
        "hub_url": "https://huggingface.co/wan-ai/wan2.1-t2v-1.3b-diffusers",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by Wan-AI",
          "ref": "Wan-AI/Wan2.1-T2V-1.3B-Diffusers",
          "engine": "diffusers",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-22T08:04:30.830Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/wan-2/t2v-1-3b/wan-ai-bf16-diffusers",
          "hub_url": "https://huggingface.co/wan-ai/wan2.1-t2v-1.3b-diffusers",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Wan-AI/Wan2.1-T2V-1.3B-Diffusers"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Tiny SD",
      "made_by": "segmind",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6335d72738e33ddc26527332/dMCE82gVz832ew-un3pqa.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Image generation",
      "community_build": false,
      "page": "/inferences/tiny-sd/base",
      "hub_url": "https://huggingface.co/segmind/tiny-sd",
      "download_gb": 1,
      "params_b": 0.32,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "makes images"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": "12th Gen Intel Core i3-12100",
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": "Intel(R) UHD Graphics 730",
          "gpu_count": 1,
          "vram_gb": 2,
          "host": "own machine",
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
        },
        "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by segmind",
        "ref": "segmind/tiny-sd",
        "engine": "generation",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-01T08:09:10.266Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/tiny-sd/base/segmind-fp32-diffusers",
        "hub_url": "https://huggingface.co/segmind/tiny-sd",
        "weeks": [
          "2026-W35",
          "2026-W36",
          "2026-W38",
          "2026-W40",
          "2026-W42",
          "2026-W43",
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 8,
      "machines_total": 8,
      "weeks_measured": [
        "2026-W35",
        "2026-W36",
        "2026-W38",
        "2026-W40",
        "2026-W42",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by segmind",
          "ref": "segmind/tiny-sd",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-01T08:09:10.266Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/tiny-sd/base/segmind-fp32-diffusers",
          "hub_url": "https://huggingface.co/segmind/tiny-sd",
          "weeks": [
            "2026-W35",
            "2026-W36",
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by segmind",
          "ref": "segmind/tiny-sd",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-14T03:57:27.249Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/tiny-sd/base/segmind-fp32-diffusers",
          "hub_url": "https://huggingface.co/segmind/tiny-sd",
          "weeks": [
            "2026-W35",
            "2026-W36",
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by segmind",
          "ref": "segmind/tiny-sd",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-16T07:37:46.439Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/tiny-sd/base/segmind-fp32-diffusers",
          "hub_url": "https://huggingface.co/segmind/tiny-sd",
          "weeks": [
            "2026-W35",
            "2026-W36",
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by segmind",
          "ref": "segmind/tiny-sd",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-22T06:08:22.501Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/tiny-sd/base/segmind-fp32-diffusers",
          "hub_url": "https://huggingface.co/segmind/tiny-sd",
          "weeks": [
            "2026-W35",
            "2026-W36",
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by segmind",
          "ref": "segmind/tiny-sd",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T04:55:30.110Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/tiny-sd/base/segmind-fp32-diffusers",
          "hub_url": "https://huggingface.co/segmind/tiny-sd",
          "weeks": [
            "2026-W35",
            "2026-W36",
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by segmind",
          "ref": "segmind/tiny-sd",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-13T23:00:24.897Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/tiny-sd/base/segmind-fp32-diffusers",
          "hub_url": "https://huggingface.co/segmind/tiny-sd",
          "weeks": [
            "2026-W35",
            "2026-W36",
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by segmind",
          "ref": "segmind/tiny-sd",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T21:25:14.031Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/tiny-sd/base/segmind-fp32-diffusers",
          "hub_url": "https://huggingface.co/segmind/tiny-sd",
          "weeks": [
            "2026-W35",
            "2026-W36",
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by segmind",
          "ref": "segmind/tiny-sd",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T13:38:52.239Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/tiny-sd/base/segmind-fp32-diffusers",
          "hub_url": "https://huggingface.co/segmind/tiny-sd",
          "weeks": [
            "2026-W35",
            "2026-W36",
            "2026-W38",
            "2026-W40",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CreativeML OpenRAIL-M",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/segmind/tiny-sd"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 8 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Whisper large-v3",
      "made_by": "openai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68783facef79a05727260de3/UPX5RQxiPGA-ZbBmArIKq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Speech to text",
      "community_build": false,
      "page": "/inferences/whisper/large-v3",
      "hub_url": "https://huggingface.co/openai/whisper-large-v3",
      "download_gb": 2.9,
      "params_b": 1.5,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "transcribes"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Marathi",
        "Nepali",
        "Punjabi",
        "Tamil",
        "Urdu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP16 (16-bit, full precision) by openai",
        "ref": "openai/whisper-large-v3",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-03T22:59:49.480Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/whisper/large-v3/openai-fp16-speech-hf",
        "hub_url": "https://huggingface.co/openai/whisper-large-v3",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by openai",
          "ref": "openai/whisper-large-v3",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T22:59:49.480Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper/large-v3/openai-fp16-speech-hf",
          "hub_url": "https://huggingface.co/openai/whisper-large-v3",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openai/whisper-large-v3"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Whisper large-v3-turbo",
      "made_by": "openai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68783facef79a05727260de3/UPX5RQxiPGA-ZbBmArIKq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Speech to text",
      "community_build": false,
      "page": "/inferences/whisper/large-v3-turbo",
      "hub_url": "https://huggingface.co/openai/whisper-large-v3-turbo",
      "download_gb": 1.5,
      "params_b": 0.81,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "transcribes"
      ],
      "languages_passed": [
        "Bengali",
        "English",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Marathi",
        "Nepali",
        "Tamil",
        "Telugu",
        "Urdu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP16 (16-bit, full precision) by openai",
        "ref": "openai/whisper-large-v3-turbo",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-21T23:40:50.590Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/whisper/large-v3-turbo/openai-fp16-speech-hf",
        "hub_url": "https://huggingface.co/openai/whisper-large-v3-turbo",
        "weeks": [
          "2026-W43",
          "2026-W45"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 5,
      "machines_total": 6,
      "weeks_measured": [
        "2026-W43",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by openai",
          "ref": "openai/whisper-large-v3-turbo",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-21T23:40:50.590Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper/large-v3-turbo/openai-fp16-speech-hf",
          "hub_url": "https://huggingface.co/openai/whisper-large-v3-turbo",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by openai",
          "ref": "openai/whisper-large-v3-turbo",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T20:01:26.298Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper/large-v3-turbo/openai-fp16-speech-hf",
          "hub_url": "https://huggingface.co/openai/whisper-large-v3-turbo",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by openai",
          "ref": "openai/whisper-large-v3-turbo",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T07:55:25.054Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper/large-v3-turbo/openai-fp16-speech-hf",
          "hub_url": "https://huggingface.co/openai/whisper-large-v3-turbo",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Mac · Apple M1 · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple M1 · 8 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.6.2",
            "cpus": 8,
            "cpu_model": "Apple M1",
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": "Apple M1",
            "gpu_count": 1,
            "vram_gb": null,
            "host": "own machine",
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|8|16|apple|0|Apple M1|1|own machine"
          },
          "machine_class": "darwin|arm64|8|16|apple|0|Apple M1|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by openai",
          "ref": "openai/whisper-large-v3-turbo",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Apple M1",
              "model_raw": "Apple M1",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": 8,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2002,
              "multi_core_hashes_per_sec": 13965.2,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-14T08:03:47.425Z",
              "measured_in": "AM_0002",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 27.7,
              "latency_ns": 7,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-14T08:03:47.969Z",
              "measured_in": "AM_0002",
              "signed": true
            },
            "storage": {
              "read_mb_s": 6263.3,
              "write_mb_s": 1266.5,
              "random_read_iops": 1295705,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-14T08:03:48.249Z",
              "measured_in": "AM_0002",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T05:06:57.448Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper/large-v3-turbo/openai-fp16-speech-hf",
          "hub_url": "https://huggingface.co/openai/whisper-large-v3-turbo",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "a5b1da"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "AMD EPYC 7B12",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 2,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP16 (16-bit, full precision) by openai",
          "ref": "openai/whisper-large-v3-turbo",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-05T08:07:50.247Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper/large-v3-turbo/openai-fp16-speech-hf",
          "hub_url": "https://huggingface.co/openai/whisper-large-v3-turbo",
          "weeks": [
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "662a4e",
            "4e94b2"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openai/whisper-large-v3-turbo"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 5 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Whisper medium",
      "made_by": "openai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68783facef79a05727260de3/UPX5RQxiPGA-ZbBmArIKq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Speech to text",
      "community_build": false,
      "page": "/inferences/whisper/medium",
      "hub_url": "https://huggingface.co/openai/whisper-medium",
      "download_gb": 2.8,
      "params_b": 0.77,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "transcribes"
      ],
      "languages_passed": [
        "English",
        "Hindi",
        "Marathi",
        "Tamil"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": "12th Gen Intel Core i3-12100",
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": "Intel(R) UHD Graphics 730",
          "gpu_count": 1,
          "vram_gb": 2,
          "host": "own machine",
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
        },
        "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by openai",
        "ref": "openai/whisper-medium",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-28T08:25:11.444Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/whisper/medium/openai-fp32-speech-hf",
        "hub_url": "https://huggingface.co/openai/whisper-medium",
        "weeks": [
          "2026-W44",
          "2026-W45"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 2,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W44",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by openai",
          "ref": "openai/whisper-medium",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-28T08:25:11.444Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper/medium/openai-fp32-speech-hf",
          "hub_url": "https://huggingface.co/openai/whisper-medium",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by openai",
          "ref": "openai/whisper-medium",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T22:38:18.890Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper/medium/openai-fp32-speech-hf",
          "hub_url": "https://huggingface.co/openai/whisper-medium",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openai/whisper-medium"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 2 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Whisper small",
      "made_by": "openai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68783facef79a05727260de3/UPX5RQxiPGA-ZbBmArIKq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Speech to text",
      "community_build": false,
      "page": "/inferences/whisper/small",
      "hub_url": "https://huggingface.co/openai/whisper-small",
      "download_gb": 0.9,
      "params_b": 0.24,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "transcribes"
      ],
      "languages_passed": [
        "English",
        "Hindi",
        "Marathi",
        "Nepali",
        "Tamil",
        "Urdu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by openai",
        "ref": "openai/whisper-small",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T11:10:31.689Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/whisper/small/openai-fp32-speech-hf",
        "hub_url": "https://huggingface.co/openai/whisper-small",
        "weeks": [
          "2026-W43",
          "2026-W44",
          "2026-W45"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 3,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W43",
        "2026-W44",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by openai",
          "ref": "openai/whisper-small",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T11:10:31.689Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper/small/openai-fp32-speech-hf",
          "hub_url": "https://huggingface.co/openai/whisper-small",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "e3654f"
          ]
        },
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by openai",
          "ref": "openai/whisper-small",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T01:22:14.407Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper/small/openai-fp32-speech-hf",
          "hub_url": "https://huggingface.co/openai/whisper-small",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by openai",
          "ref": "openai/whisper-small",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T22:35:28.607Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper/small/openai-fp32-speech-hf",
          "hub_url": "https://huggingface.co/openai/whisper-small",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openai/whisper-small"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 3 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Whisper small hindi (vasista22)",
      "made_by": "vasista22",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68783facef79a05727260de3/UPX5RQxiPGA-ZbBmArIKq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Speech to text",
      "community_build": false,
      "page": "/inferences/whisper-indic-ft/hi-small",
      "hub_url": "https://huggingface.co/openai/whisper-small",
      "download_gb": 0.9,
      "params_b": 0.24,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "transcribes"
      ],
      "languages_passed": [
        "Hindi"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by vasista22",
        "ref": "vasista22/whisper-hindi-small",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T11:33:45.752Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/whisper-indic-ft/hi-small/vasista22-fp32-speech-hf",
        "hub_url": "https://huggingface.co/vasista22/whisper-hindi-small",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by vasista22",
          "ref": "vasista22/whisper-hindi-small",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T11:33:45.752Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper-indic-ft/hi-small/vasista22-fp32-speech-hf",
          "hub_url": "https://huggingface.co/vasista22/whisper-hindi-small",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openai/whisper-small"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Whisper small kannada (vasista22)",
      "made_by": "vasista22",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68783facef79a05727260de3/UPX5RQxiPGA-ZbBmArIKq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Speech to text",
      "community_build": false,
      "page": "/inferences/whisper-indic-ft/kn-small",
      "hub_url": "https://huggingface.co/openai/whisper-small",
      "download_gb": 0.9,
      "params_b": 0.24,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "transcribes"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by vasista22",
        "ref": "vasista22/whisper-kannada-small",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 181,
            "multi_core_hashes_per_sec": 436.8,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-30T20:43:55.306Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 3.4,
            "latency_ns": 49.1,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-30T20:43:56.203Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 4271.7,
            "write_mb_s": 177.9,
            "random_read_iops": 391634,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T20:43:56.799Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T18:01:21.705Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/whisper-indic-ft/kn-small/vasista22-fp32-speech-hf",
        "hub_url": "https://huggingface.co/vasista22/whisper-kannada-small",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by vasista22",
          "ref": "vasista22/whisper-kannada-small",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T18:01:21.705Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper-indic-ft/kn-small/vasista22-fp32-speech-hf",
          "hub_url": "https://huggingface.co/vasista22/whisper-kannada-small",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openai/whisper-small"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Whisper small tamil (vasista22)",
      "made_by": "vasista22",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68783facef79a05727260de3/UPX5RQxiPGA-ZbBmArIKq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Speech to text",
      "community_build": false,
      "page": "/inferences/whisper-indic-ft/ta-small",
      "hub_url": "https://huggingface.co/openai/whisper-small",
      "download_gb": 0.9,
      "params_b": 0.24,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "transcribes"
      ],
      "languages_passed": [
        "Tamil"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by vasista22",
        "ref": "vasista22/whisper-tamil-small",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:01:05.273Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/whisper-indic-ft/ta-small/vasista22-fp32-speech-hf",
        "hub_url": "https://huggingface.co/vasista22/whisper-tamil-small",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by vasista22",
          "ref": "vasista22/whisper-tamil-small",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:01:05.273Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper-indic-ft/ta-small/vasista22-fp32-speech-hf",
          "hub_url": "https://huggingface.co/vasista22/whisper-tamil-small",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openai/whisper-small"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Whisper small telugu (vasista22)",
      "made_by": "vasista22",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68783facef79a05727260de3/UPX5RQxiPGA-ZbBmArIKq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Speech to text",
      "community_build": false,
      "page": "/inferences/whisper-indic-ft/te-small",
      "hub_url": "https://huggingface.co/openai/whisper-small",
      "download_gb": 0.9,
      "params_b": 0.24,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "transcribes"
      ],
      "languages_passed": [
        "Telugu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by vasista22",
        "ref": "vasista22/whisper-telugu-small",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-26T21:17:15.583Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/whisper-indic-ft/te-small/vasista22-fp32-speech-hf",
        "hub_url": "https://huggingface.co/vasista22/whisper-telugu-small",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by vasista22",
          "ref": "vasista22/whisper-telugu-small",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-26T21:17:15.583Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/whisper-indic-ft/te-small/vasista22-fp32-speech-hf",
          "hub_url": "https://huggingface.co/vasista22/whisper-telugu-small",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openai/whisper-small"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Xtts-v2 (17 languages, 6-second clone)",
      "made_by": "coqui",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1648459271425-61dbf25b99eac69222659467.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text to speech",
      "community_build": false,
      "page": "/inferences/xtts/v2",
      "hub_url": "https://huggingface.co/coqui/XTTS-v2",
      "download_gb": 1.9,
      "params_b": 0.47,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "speaks"
      ],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by coqui",
        "ref": "coqui/XTTS-v2",
        "engine": "speech-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-06T17:14:33.298Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/xtts/v2/coqui-fp32-speech-hf",
        "hub_url": "https://huggingface.co/coqui/xtts-v2",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by coqui",
          "ref": "coqui/XTTS-v2",
          "engine": "speech-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T17:14:33.298Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/xtts/v2/coqui-fp32-speech-hf",
          "hub_url": "https://huggingface.co/coqui/xtts-v2",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Coqui Public Model License licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/coqui/XTTS-v2"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Hunyuan MT 2 1.8B",
      "made_by": "tencent",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/Lp3m-XLpjQGwBItlvn69q.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/hy-mt-2/1.8b",
      "hub_url": "https://huggingface.co/tencent/Hy-MT2-1.8B",
      "download_gb": 0.9,
      "params_b": 1.8,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": "AMD Ryzen 7 5800X 8-Core",
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "GeForce RTX 3080 Ti",
          "gpu_count": 1,
          "vram_gb": 12,
          "host": "own machine",
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
        "machines": 1,
        "tokens_per_sec": 221.76,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by tencent",
        "ref": "hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": 1.1595,
        "cost_breakdown": {
          "tokens_per_sec": 221.76,
          "usd_per_hour": 0.9257,
          "usd_per_million": 1.1595,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "15 of 16 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 150,
              "usd_per_million": 0.6764
            },
            {
              "resource": "Memory",
              "quantity": "1.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.879,
              "usd_per_million": 0.004
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.4509
            },
            {
              "resource": "Energy",
              "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 6.25,
              "usd_per_million": 0.0282
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-10T11:20:52.995Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "S5f/XRcwG7fTN1OlHLHiODerHjAf+JqE7SvC+MJE8yYYEv/ySr4ui5T6Zx7ucagCPSuAeTsojhpMsDJc+S9+CQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "pCzj+fIjZyfPuWAP9hghUgs88J6II22ffnsq0T8FTew="
        },
        "build_page": "/inferences/hy-mt-2/1.8b/tencent-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/tencent/hy-mt2-1.8b-gguf",
        "weeks": [
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "cheapest": {
        "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": "AMD Ryzen 7 5800X 8-Core",
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "GeForce RTX 3080 Ti",
          "gpu_count": 1,
          "vram_gb": 12,
          "host": "own machine",
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
        "machines": 1,
        "tokens_per_sec": 221.76,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by tencent",
        "ref": "hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": 1.1595,
        "cost_breakdown": {
          "tokens_per_sec": 221.76,
          "usd_per_hour": 0.9257,
          "usd_per_million": 1.1595,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "15 of 16 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 150,
              "usd_per_million": 0.6764
            },
            {
              "resource": "Memory",
              "quantity": "1.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.879,
              "usd_per_million": 0.004
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 0.4509
            },
            {
              "resource": "Energy",
              "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 6.25,
              "usd_per_million": 0.0282
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-10T11:20:52.995Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
          "signature": "S5f/XRcwG7fTN1OlHLHiODerHjAf+JqE7SvC+MJE8yYYEv/ySr4ui5T6Zx7ucagCPSuAeTsojhpMsDJc+S9+CQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "pCzj+fIjZyfPuWAP9hghUgs88J6II22ffnsq0T8FTew="
        },
        "build_page": "/inferences/hy-mt-2/1.8b/tencent-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/tencent/hy-mt2-1.8b-gguf",
        "weeks": [
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "machine_kinds": 4,
      "machines_total": 5,
      "weeks_measured": [
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": 221.76,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by tencent",
          "ref": "hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 1.1595,
          "cost_breakdown": {
            "tokens_per_sec": 221.76,
            "usd_per_hour": 0.9257,
            "usd_per_million": 1.1595,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "15 of 16 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 150,
                "usd_per_million": 0.6764
              },
              {
                "resource": "Memory",
                "quantity": "1.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.879,
                "usd_per_million": 0.004
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 0.4509
              },
              {
                "resource": "Energy",
                "quantity": "225 W modelled (15 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 6.25,
                "usd_per_million": 0.0282
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T11:20:52.995Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220bc650b59af8794896031d2e82bca0e119f390e21d690963ae47578238084bf54",
            "signature": "S5f/XRcwG7fTN1OlHLHiODerHjAf+JqE7SvC+MJE8yYYEv/ySr4ui5T6Zx7ucagCPSuAeTsojhpMsDJc+S9+CQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "pCzj+fIjZyfPuWAP9hghUgs88J6II22ffnsq0T8FTew="
          },
          "build_page": "/inferences/hy-mt-2/1.8b/tencent-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/tencent/hy-mt2-1.8b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 36.5,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by tencent",
          "ref": "hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 5.9337,
          "cost_breakdown": {
            "tokens_per_sec": 36.5,
            "usd_per_hour": 0.7797,
            "usd_per_million": 5.9337,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 3.0138
              },
              {
                "resource": "Memory",
                "quantity": "1.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.879,
                "usd_per_million": 0.0241
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.7398
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.156
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "yTk/m+YOLv42J+26aZq/6M0CjCc/pO0gbVKFx5wvGlslvYMo8lgnDFEChtwwjHqrGO918YCfIzOELKW4vNObAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "99PE4dNrYcS6W3TTiS67PWoe2oOgZEF1NgRU6Ui7+r4="
          },
          "build_page": "/inferences/hy-mt-2/1.8b/tencent-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/tencent/hy-mt2-1.8b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 15.31,
          "context_tokens": null,
          "build": "Q3_K_S (3-bit) by mradermacher",
          "ref": "hf.co/mradermacher/Hy-MT2-1.8B-GGUF:Q3_K_S",
          "engine": "ollama",
          "cost_usd_per_million": 4.6786,
          "cost_breakdown": {
            "tokens_per_sec": 15.31,
            "usd_per_hour": 0.2579,
            "usd_per_million": 4.6786,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 4.5709
              },
              {
                "resource": "Memory",
                "quantity": "0.8 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.676,
                "usd_per_million": 0.0441
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0635
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "/ujE9QwdaKcDrlf7aFO5HlWFJaxhotaa3I7uS8wP6qQ0a5JZfbGznK76W/3h376YNiR2h06JxCDzjIQ124J7Cw==",
            "signing_payload_b64": "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",
            "payload_sha256": "zcN+6TqviVRG5jGJ0pkRxVAuaM9oZSplJH7Uv8JqMcc="
          },
          "build_page": "/inferences/hy-mt-2/1.8b/mradermacher-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/mradermacher/hy-mt2-1.8b-gguf",
          "weeks": [
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 14.17,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by tencent",
          "ref": "hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 5.0706,
          "cost_breakdown": {
            "tokens_per_sec": 14.17,
            "usd_per_hour": 0.2587,
            "usd_per_million": 5.0706,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 4.94
              },
              {
                "resource": "Memory",
                "quantity": "1.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.879,
                "usd_per_million": 0.062
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0686
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "Cghz/vOSgpMIIJTmK4/7JxW09UynwtcpE1/7PM5nruGcm4x3wLJQViHt6iwbWriJ2sLLyFxbg0qZl0d9QmaVBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "j+HXRQHEjnhv0tvpkFb5wLZ8N8ZnG7pzUj+wDrVNEUQ="
          },
          "build_page": "/inferences/hy-mt-2/1.8b/tencent-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/tencent/hy-mt2-1.8b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": 10.63,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by tencent",
          "ref": "hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 6.7573,
          "cost_breakdown": {
            "tokens_per_sec": 10.63,
            "usd_per_hour": 0.2587,
            "usd_per_million": 6.7573,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 6.5832
              },
              {
                "resource": "Memory",
                "quantity": "1.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.879,
                "usd_per_million": 0.0827
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.0914
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:46:12.289Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112203fe69ce44d7b1418dd426331a539f12c1de428a235017442d41ceab0a29e3c79",
            "signature": "1I3J15zrqtK2sw2NQWhjpvF13Q1dcC0YPfCFK2FCEdDTvcB0XNrgKgz4DNW/w7qv7tEevciL6xYWVJO2mxFRBA==",
            "signing_payload_b64": "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",
            "payload_sha256": "38UlxBuYq+4zsiivvyPkH/7oM2pcA/XDsdw6zV3KGRM="
          },
          "build_page": "/inferences/hy-mt-2/1.8b/tencent-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/tencent/hy-mt2-1.8b-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "tencent/hy-mt2-1.8b",
        "url": "https://openrouter.ai/tencent/hy-mt2-1.8b",
        "input_usd_per_million": 0.044,
        "output_usd_per_million": 0.177,
        "free_tier": false,
        "context_tokens": 8192,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Tencent",
            "tag": "tencent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.044,
            "usd_per_m_out": 0.177,
            "context_length": 8192,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/tencent/Hy-MT2-1.8B"
      },
      "builds": 6,
      "variants": [],
      "bottom_line": "222 tokens a second on Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM. About $1.16 per million tokens here, against OpenRouter's $0.18: the API is 6.6× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "Hunyuan MT 1.5 1.8B",
      "made_by": "tencent",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/Lp3m-XLpjQGwBItlvn69q.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/hy-mt-1-5/1.8b",
      "hub_url": "https://huggingface.co/tencent/HY-MT1.5-1.8B-GGUF",
      "download_gb": 1.1,
      "params_b": 1.8,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 46.49,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by tencent",
        "ref": "hf.co/tencent/HY-MT1.5-1.8B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 4.6588,
        "cost_breakdown": {
          "tokens_per_sec": 46.49,
          "usd_per_hour": 0.7797,
          "usd_per_million": 4.6588,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 2.3663
            },
            {
              "resource": "Memory",
              "quantity": "1.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.879,
              "usd_per_million": 0.0189
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 2.1512
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.1225
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "NNbTmCLwHGAorvh1JM04OX7kNiqaUDV9mpXoUKND7kbAu92PwN3OX5bpfGDlDetBsra/lb1h3y0u8tDqLxjeDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "i6nuKjoVXCDaeuyXdIRf9fTeYHlJPwAxundLabnF1XY="
        },
        "build_page": "/inferences/hy-mt-1-5/1.8b/tencent-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/tencent/hy-mt1.5-1.8b-gguf",
        "weeks": [
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 46.49,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by tencent",
        "ref": "hf.co/tencent/HY-MT1.5-1.8B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 4.6588,
        "cost_breakdown": {
          "tokens_per_sec": 46.49,
          "usd_per_hour": 0.7797,
          "usd_per_million": 4.6588,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 2.3663
            },
            {
              "resource": "Memory",
              "quantity": "1.1 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.879,
              "usd_per_million": 0.0189
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 2.1512
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.1225
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "NNbTmCLwHGAorvh1JM04OX7kNiqaUDV9mpXoUKND7kbAu92PwN3OX5bpfGDlDetBsra/lb1h3y0u8tDqLxjeDA==",
          "signing_payload_b64": "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",
          "payload_sha256": "i6nuKjoVXCDaeuyXdIRf9fTeYHlJPwAxundLabnF1XY="
        },
        "build_page": "/inferences/hy-mt-1-5/1.8b/tencent-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/tencent/hy-mt1.5-1.8b-gguf",
        "weeks": [
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 2,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 46.49,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by tencent",
          "ref": "hf.co/tencent/HY-MT1.5-1.8B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 4.6588,
          "cost_breakdown": {
            "tokens_per_sec": 46.49,
            "usd_per_hour": 0.7797,
            "usd_per_million": 4.6588,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 2.3663
              },
              {
                "resource": "Memory",
                "quantity": "1.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.879,
                "usd_per_million": 0.0189
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 2.1512
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.1225
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "NNbTmCLwHGAorvh1JM04OX7kNiqaUDV9mpXoUKND7kbAu92PwN3OX5bpfGDlDetBsra/lb1h3y0u8tDqLxjeDA==",
            "signing_payload_b64": "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",
            "payload_sha256": "i6nuKjoVXCDaeuyXdIRf9fTeYHlJPwAxundLabnF1XY="
          },
          "build_page": "/inferences/hy-mt-1-5/1.8b/tencent-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/tencent/hy-mt1.5-1.8b-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": 7.41,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by tencent",
          "ref": "hf.co/tencent/HY-MT1.5-1.8B-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": 9.6986,
          "cost_breakdown": {
            "tokens_per_sec": 7.41,
            "usd_per_hour": 0.2587,
            "usd_per_million": 9.6986,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "7 of 8 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 70,
                "usd_per_million": 9.4488
              },
              {
                "resource": "Memory",
                "quantity": "1.1 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.879,
                "usd_per_million": 0.1186
              },
              {
                "resource": "Energy",
                "quantity": "35 W modelled (7 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.972,
                "usd_per_million": 0.1312
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T14:04:06.882Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112201536df94f88c2f2dce9c87297f42e0902d5d635ea2736f99141c3ebcea2e5c02",
            "signature": "AL8OpM9y/9FbdJu2l6PLLBZKKS3fwAnSI/ySdbvmJ4hQbrVYqvTHKliA0r2AL3yHLzvQW75vpqcJR2/mofNbDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "QAui5ZrGmqP3ObE5U+lcBKsLmWgW6Nte9XTifeNVPUI="
          },
          "build_page": "/inferences/hy-mt-1-5/1.8b/tencent-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/tencent/hy-mt1.5-1.8b-gguf",
          "weeks": [
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "Q4_0 (4-bit) by mradermacher",
          "ref": "hf.co/mradermacher/HY-MT1.5-1.8B-i1-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-03T21:20:41.123Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/hy-mt-1-5/1.8b~mradermacher-i1/mradermacher-q4_0-ollama",
          "hub_url": "https://huggingface.co/mradermacher/hy-mt1.5-1.8b-i1-gguf",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": "https://huggingface.co/tencent/HY-MT1.5-1.8B-GGUF"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "46 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Home-FunctionGemma-270m",
      "made_by": "acon96",
      "maker_logo": "https://huggingface.co/avatars/e784abe90576ef7ea4544938b4e9e6ff.svg",
      "based_on": "FunctionGemma 270M",
      "variant_of": "/inferences/functiongemma/270m",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/functiongemma/270m~acon96-home",
      "hub_url": "https://huggingface.co/acon96/Home-FunctionGemma-270m",
      "download_gb": 0.2,
      "params_b": 0.27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 35.88,
        "context_tokens": null,
        "build": "Q3_K_S (3-bit) by mradermacher",
        "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q3_K_S",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.8528,
        "cost_breakdown": {
          "tokens_per_sec": 35.88,
          "usd_per_hour": 0.1102,
          "usd_per_million": 0.8528,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.8361
            },
            {
              "resource": "Memory",
              "quantity": "0.2 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.183,
              "usd_per_million": 0.0051
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0116
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-25T07:41:14.640Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
          "signature": "tzbpkEe8EG6WTXyAokn+JdJAn7jzS6fBXLSuV0TfhtY/Dio6v6pJREuBYqfT9AbugH5leQalYkRAE4WSNmgoCg==",
          "signing_payload_b64": "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",
          "payload_sha256": "XZyv4iHicMADbGT3gbg1ck/XYMKGk0y2YXEQujqstsM="
        },
        "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-q3_k_s-ollama",
        "hub_url": "https://huggingface.co/mradermacher/home-functiongemma-270m-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "acbe75"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": 35.88,
        "context_tokens": null,
        "build": "Q3_K_S (3-bit) by mradermacher",
        "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q3_K_S",
        "engine": "ollama-host",
        "cost_usd_per_million": 0.8528,
        "cost_breakdown": {
          "tokens_per_sec": 35.88,
          "usd_per_hour": 0.1102,
          "usd_per_million": 0.8528,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 0.8361
            },
            {
              "resource": "Memory",
              "quantity": "0.2 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 0.183,
              "usd_per_million": 0.0051
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0116
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-25T07:41:14.640Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
          "signature": "tzbpkEe8EG6WTXyAokn+JdJAn7jzS6fBXLSuV0TfhtY/Dio6v6pJREuBYqfT9AbugH5leQalYkRAE4WSNmgoCg==",
          "signing_payload_b64": "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",
          "payload_sha256": "XZyv4iHicMADbGT3gbg1ck/XYMKGk0y2YXEQujqstsM="
        },
        "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-q3_k_s-ollama",
        "hub_url": "https://huggingface.co/mradermacher/home-functiongemma-270m-gguf",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "acbe75"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 7,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 35.88,
          "context_tokens": null,
          "build": "Q3_K_S (3-bit) by mradermacher",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.8528,
          "cost_breakdown": {
            "tokens_per_sec": 35.88,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.8528,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8361
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.183,
                "usd_per_million": 0.0051
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0116
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "tzbpkEe8EG6WTXyAokn+JdJAn7jzS6fBXLSuV0TfhtY/Dio6v6pJREuBYqfT9AbugH5leQalYkRAE4WSNmgoCg==",
            "signing_payload_b64": "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",
            "payload_sha256": "XZyv4iHicMADbGT3gbg1ck/XYMKGk0y2YXEQujqstsM="
          },
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/mradermacher/home-functiongemma-270m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 34.81,
          "context_tokens": null,
          "build": "IQ4_XS (4-bit) by mradermacher",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:IQ4_XS",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.8793,
          "cost_breakdown": {
            "tokens_per_sec": 34.81,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.8793,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.8619
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.187,
                "usd_per_million": 0.0054
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.012
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "O0nKhtOFxpfIpya9pFEDmJW3HP3sRYRUoUqyRHrxwKwXbxi7OjWpWPZnhDyd2cLiu70atOePegxrEZaFL4avDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "lF352rhXqEF9b9MPhqiULX+B+ooFb2rDscp0zDP07hc="
          },
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-iq4_xs-ollama",
          "hub_url": "https://huggingface.co/mradermacher/home-functiongemma-270m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 34.52,
          "context_tokens": null,
          "build": "Q2_K (2-bit) by mradermacher",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q2_K",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.8864,
          "cost_breakdown": {
            "tokens_per_sec": 34.52,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.8864,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.869
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.184,
                "usd_per_million": 0.0053
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0121
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "NrQzePjdOBv88WWNWKAzYKVTos1rauqZ0NktiZPcu6vFH9gn0I2lsIvrQXVP+lbqKsV28VpHm8i24JZ8SVBfBw==",
            "signing_payload_b64": "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",
            "payload_sha256": "6h1i0nRLzcSte/na3Tw3ePx7lxC85ruqzVHAOJpmQVU="
          },
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-q2_k-ollama",
          "hub_url": "https://huggingface.co/mradermacher/home-functiongemma-270m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 33.28,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by mradermacher",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9197,
          "cost_breakdown": {
            "tokens_per_sec": 33.28,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.9197,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.9013
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.196,
                "usd_per_million": 0.0059
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0125
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "lkMyeXX1dWklFUlCSoYjkEUhPO+SUsO04bfFn1SB6p6GzmHNlqFBkVhTiny+X1GSr0o0lO806VXQrebsRT+zAQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "lKPUcZUSmw9v3Bzid6+RRGPQEEmr3A6NveVI+7EmLFs="
          },
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/mradermacher/home-functiongemma-270m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 32.8,
          "context_tokens": null,
          "build": "Q4_K_S (4-bit) by mradermacher",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q4_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9333,
          "cost_breakdown": {
            "tokens_per_sec": 32.8,
            "usd_per_hour": 0.1102,
            "usd_per_million": 0.9333,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.9146
              },
              {
                "resource": "Memory",
                "quantity": "0.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.194,
                "usd_per_million": 0.0059
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0127
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "cjHtAisgoqrEm3tsu0iqmZobRO944Vr+tc7/A+Davy13qoW2FmgSjyojOHQyAFdF+rx/BUej3wIzBBbgEHBjCQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "4d76KNwnw82us4h95HUo/V2x8HrqqK9SB0xxd9fGGng="
          },
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-q4_k_s-ollama",
          "hub_url": "https://huggingface.co/mradermacher/home-functiongemma-270m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 31.99,
          "context_tokens": null,
          "build": "Q8_0 (8-bit) by mradermacher",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 0.9578,
          "cost_breakdown": {
            "tokens_per_sec": 31.99,
            "usd_per_hour": 0.1103,
            "usd_per_million": 0.9578,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 0.9377
              },
              {
                "resource": "Memory",
                "quantity": "0.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.226,
                "usd_per_million": 0.0071
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.013
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "JtIK2OreU3Sx4Z8uESr0dtJ6iCbJAsitkamJ5Vux8D+zvHTdU/4nIfEvxCWfVyKIQPfSqPURW6wae50Ge+S5BA==",
            "signing_payload_b64": "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",
            "payload_sha256": "ivGWW/hE5ieGh2DOMOyrPp+U30SzJEdCR5+GeKDIJVM="
          },
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-q8_0-ollama",
          "hub_url": "https://huggingface.co/mradermacher/home-functiongemma-270m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": 22.54,
          "context_tokens": null,
          "build": "F16 (16-bit, full precision) by mradermacher",
          "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:F16",
          "engine": "ollama-host",
          "cost_usd_per_million": 1.3679,
          "cost_breakdown": {
            "tokens_per_sec": 22.54,
            "usd_per_hour": 0.111,
            "usd_per_million": 1.3679,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 1.3307
              },
              {
                "resource": "Memory",
                "quantity": "0.5 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 0.421,
                "usd_per_million": 0.0187
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0185
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-25T07:41:14.640Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220267eb5f89d88bfed1fb33798c96d6ea96399f390cc6815baec7d9cb177acbe75",
            "signature": "Wx2jZyk3lqJbZmqy9a67YBKi2qqjNJhIFO/LAx/MVWLd7E+0K3RkKmUQ6bQsgUmeFyxt7VpfxtKilDd5+iLLBA==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6MjIuNTQ0NzE5MDA3MjYzNzE2LCJ1bml0IjoidG9rZW5zL3MiLCJ0cyI6MTc5MjkxNDA3NDY0MDg5MjYyNywibm9kZV9kaWQiOiJkaWQ6ZXBuOjAwMjQwODAxMTIyMDI2N2ViNWY4OWQ4OGJmZWQxZmIzMzc5OGM5NmQ2ZWE5NjM5OWYzOTBjYzY4MTViYWVjN2Q5Y2IxNzdhY2JlNzUiLCJleHRyYSI6eyJtb2RlbCI6ImhmLmNvL21yYWRlcm1hY2hlci9Ib21lLUZ1bmN0aW9uR2VtbWEtMjcwbS1HR1VGOkYxNiIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo1NCwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjo2MDMuMDY4MDYxOTk3OTk5OCwidG90YWxfdG9rZW5zIjoxMzU5Nn19",
            "payload_sha256": "Ab89laejN8zQQKUJG/qSyjOL7BzD594fJHx604XpeNs="
          },
          "build_page": "/inferences/functiongemma/270m~acon96-home/mradermacher-f16-ollama",
          "hub_url": "https://huggingface.co/mradermacher/home-functiongemma-270m-gguf",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "acbe75"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Gemma terms of use",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/acon96/Home-FunctionGemma-270m"
      },
      "builds": 7,
      "variants": [],
      "bottom_line": "36 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Muse Glimmer 30B",
      "made_by": "meta-models",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/64c115d1f661ace7c4022432/EbzfQisFJBEcKWDWLyV5O.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/muse-glimmer/30b",
      "hub_url": "https://huggingface.co/meta-models/Muse-Glimmer-30B",
      "download_gb": 15.6,
      "params_b": 30,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 6.9,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by meta-models",
        "ref": "hf.co/meta-models/Muse-Glimmer-30B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 33.14,
        "cost_breakdown": {
          "tokens_per_sec": 6.9,
          "usd_per_hour": 0.8233,
          "usd_per_million": 33.14,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 15.9398
            },
            {
              "resource": "Memory",
              "quantity": "15.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 13.005,
              "usd_per_million": 1.8845
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 14.4907
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.8251
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "dMDssy8soIcd+lIHE7HG9kD6lZsN72dm/zX2ExyZ59IHaBXxamFT8D/4iu4pbTICqGeh7RrvtCWHADlDPOeSDg==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6Ni45MDA5ODU1MzIxMjM0OTQsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzk0MDI4MTkyOTcwOTcyMjA1LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwNTI0MDUwZTkyOTZkMzEyY2QyODI4YzQyNjE2NWJmZjIzYmFlZWY3NjlhZGE5NzA4MzA5ODgzMmYwNmRhZTY3OSIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbnZpZGlhLzY0ZyIsIm1vZGVsIjoiaGYuY28vbWV0YS1tb2RlbHMvTXVzZS1HbGltbWVyLTMwQi1HR1VGOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50Ijo1OCwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjoyNS4yMTM3ODk5OTk5OTk5OTYsInRvdGFsX3Rva2VucyI6MTc0fX0=",
          "payload_sha256": "6ja5GGccIEW2JG1KsRCCNkZBw/mCaryhbZ7uaY17pAk="
        },
        "build_page": "/inferences/muse-glimmer/30b/meta-models-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/meta-models/muse-glimmer-30b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": 6.9,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by meta-models",
        "ref": "hf.co/meta-models/Muse-Glimmer-30B-GGUF:Q4_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 33.14,
        "cost_breakdown": {
          "tokens_per_sec": 6.9,
          "usd_per_hour": 0.8233,
          "usd_per_million": 33.14,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "11 of 12 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 110,
              "usd_per_million": 15.9398
            },
            {
              "resource": "Memory",
              "quantity": "15.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 13.005,
              "usd_per_million": 1.8845
            },
            {
              "resource": "GPU",
              "quantity": "one card",
              "rate": "$0.36 per GPU-hour",
              "uusd_per_second": 100,
              "usd_per_million": 14.4907
            },
            {
              "resource": "Energy",
              "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 5.694,
              "usd_per_million": 0.8251
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-07T05:09:52.970Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
          "signature": "dMDssy8soIcd+lIHE7HG9kD6lZsN72dm/zX2ExyZ59IHaBXxamFT8D/4iu4pbTICqGeh7RrvtCWHADlDPOeSDg==",
          "signing_payload_b64": "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",
          "payload_sha256": "6ja5GGccIEW2JG1KsRCCNkZBw/mCaryhbZ7uaY17pAk="
        },
        "build_page": "/inferences/muse-glimmer/30b/meta-models-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/meta-models/muse-glimmer-30b-gguf",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": 6.9,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by meta-models",
          "ref": "hf.co/meta-models/Muse-Glimmer-30B-GGUF:Q4_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 33.14,
          "cost_breakdown": {
            "tokens_per_sec": 6.9,
            "usd_per_hour": 0.8233,
            "usd_per_million": 33.14,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "11 of 12 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 110,
                "usd_per_million": 15.9398
              },
              {
                "resource": "Memory",
                "quantity": "15.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 13.005,
                "usd_per_million": 1.8845
              },
              {
                "resource": "GPU",
                "quantity": "one card",
                "rate": "$0.36 per GPU-hour",
                "uusd_per_second": 100,
                "usd_per_million": 14.4907
              },
              {
                "resource": "Energy",
                "quantity": "205 W modelled (11 threads × 5 W + 1 × 150 W card)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 5.694,
                "usd_per_million": 0.8251
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T05:09:52.970Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220524050e9296d312cd2828c426165bff23baeef769ada97083098832f06dae679",
            "signature": "dMDssy8soIcd+lIHE7HG9kD6lZsN72dm/zX2ExyZ59IHaBXxamFT8D/4iu4pbTICqGeh7RrvtCWHADlDPOeSDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "6ja5GGccIEW2JG1KsRCCNkZBw/mCaryhbZ7uaY17pAk="
          },
          "build_page": "/inferences/muse-glimmer/30b/meta-models-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/meta-models/muse-glimmer-30b-gguf",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "meta/muse-glimmer-30b",
        "url": "https://openrouter.ai/meta/muse-glimmer-30b",
        "input_usd_per_million": 0.3,
        "output_usd_per_million": 1.2,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.35,
            "usd_per_m_out": 1.5,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/meta-models/Muse-Glimmer-30B"
      },
      "builds": 5,
      "variants": [
        {
          "model": "Muse-Glimmer-30B-Heretic-Uncensored-GGUF",
          "made_by": "0bserverx",
          "page": "/inferences/muse-glimmer/30b~0bserverx-heretic-uncensored",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "Muse-Glimmer-30B-pruned-V2.5-en-pl-GGUF",
          "made_by": "bluevoid-pl",
          "page": "/inferences/muse-glimmer/30b~bluevoid-pl-pruned-v2-5-en",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "Muse-Glimmer-30B-VL-gptq-mlx-jang",
          "made_by": "roman220220",
          "page": "/inferences/muse-glimmer/30b~roman220220-vl-jang",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "6.9 tokens a second on Linux · L4 23 GB · 12 CPUs · 53 GB RAM. About $33 per million tokens here, against OpenRouter's $1.20: the API is 28× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
      "made_by": "DavidAU",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/65ea44635b64331c067d3751/yCim-7c3tm67o5wWP_6cE.jpeg",
      "based_on": "gpt-oss 20B",
      "variant_of": "/inferences/gpt-oss/20b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/gpt-oss/20b~davidau-openai-claude-4-5-opus-heretic-uncensored",
      "hub_url": "https://huggingface.co/DavidAU/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
      "download_gb": 11.2,
      "params_b": 20,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 4.82,
        "context_tokens": null,
        "build": "Q4_0 (4-bit) by mradermacher",
        "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q4_0",
        "engine": "ollama-host",
        "cost_usd_per_million": 8.2607,
        "cost_breakdown": {
          "tokens_per_sec": 4.82,
          "usd_per_hour": 0.1434,
          "usd_per_million": 8.2607,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 6.2196
            },
            {
              "resource": "Memory",
              "quantity": "11.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 9.428,
              "usd_per_million": 1.9546
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0865
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T06:35:24.096Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "tWuYKHXMN171qhfeC2n94lPtEPX0hy10zLDTcSbRgECJD1Wzf9lArjimkKPZSQQnEz8ni5VIrWLzenyulRc6BQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "uRLhHieWlXAf5UsGL3B7oejF/sRsQgtSGXwHmITGL+s="
        },
        "build_page": "/inferences/gpt-oss/20b~davidau-openai-claude-4-5-opus-heretic-uncensored/mradermacher-q4_0-ollama",
        "hub_url": "https://huggingface.co/mradermacher/openai-gpt-oss-20b-claude-4.5-opus-heretic-uncensored-i1-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 4.82,
        "context_tokens": null,
        "build": "Q4_0 (4-bit) by mradermacher",
        "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q4_0",
        "engine": "ollama-host",
        "cost_usd_per_million": 8.2607,
        "cost_breakdown": {
          "tokens_per_sec": 4.82,
          "usd_per_hour": 0.1434,
          "usd_per_million": 8.2607,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 6.2196
            },
            {
              "resource": "Memory",
              "quantity": "11.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 9.428,
              "usd_per_million": 1.9546
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0865
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T06:35:24.096Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "tWuYKHXMN171qhfeC2n94lPtEPX0hy10zLDTcSbRgECJD1Wzf9lArjimkKPZSQQnEz8ni5VIrWLzenyulRc6BQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "uRLhHieWlXAf5UsGL3B7oejF/sRsQgtSGXwHmITGL+s="
        },
        "build_page": "/inferences/gpt-oss/20b~davidau-openai-claude-4-5-opus-heretic-uncensored/mradermacher-q4_0-ollama",
        "hub_url": "https://huggingface.co/mradermacher/openai-gpt-oss-20b-claude-4.5-opus-heretic-uncensored-i1-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W44",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 4.82,
          "context_tokens": null,
          "build": "Q4_0 (4-bit) by mradermacher",
          "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q4_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 8.2607,
          "cost_breakdown": {
            "tokens_per_sec": 4.82,
            "usd_per_hour": 0.1434,
            "usd_per_million": 8.2607,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.2196
              },
              {
                "resource": "Memory",
                "quantity": "11.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 9.428,
                "usd_per_million": 1.9546
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0865
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T06:35:24.096Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "tWuYKHXMN171qhfeC2n94lPtEPX0hy10zLDTcSbRgECJD1Wzf9lArjimkKPZSQQnEz8ni5VIrWLzenyulRc6BQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "uRLhHieWlXAf5UsGL3B7oejF/sRsQgtSGXwHmITGL+s="
          },
          "build_page": "/inferences/gpt-oss/20b~davidau-openai-claude-4-5-opus-heretic-uncensored/mradermacher-q4_0-ollama",
          "hub_url": "https://huggingface.co/mradermacher/openai-gpt-oss-20b-claude-4.5-opus-heretic-uncensored-i1-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 4.72,
          "context_tokens": null,
          "build": "Q3_K_S (3-bit) by mradermacher",
          "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q3_K_S",
          "engine": "ollama-host",
          "cost_usd_per_million": 8.4252,
          "cost_breakdown": {
            "tokens_per_sec": 4.72,
            "usd_per_hour": 0.1432,
            "usd_per_million": 8.4252,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.3543
              },
              {
                "resource": "Memory",
                "quantity": "11.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 9.36,
                "usd_per_million": 1.9825
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0883
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T06:35:24.096Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "+xtVIsd6qc1eivlsb6wbHAbvpew5wzmWuw4WWotRENaG9xyUZ8sWhrfa4KCYs5CVk+sBDLZaWtTbvwFz+6gNDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "SkIjIbR7PzFwQbcV/Ng/nk8P0dAOxVdHZBYodBYhJ6k="
          },
          "build_page": "/inferences/gpt-oss/20b~davidau-openai-claude-4-5-opus-heretic-uncensored/mradermacher-q3_k_s-ollama",
          "hub_url": "https://huggingface.co/mradermacher/openai-gpt-oss-20b-claude-4.5-opus-heretic-uncensored-i1-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 4.23,
          "context_tokens": null,
          "build": "IQ1_M (1-bit) by mradermacher",
          "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:IQ1_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 9.3918,
          "cost_breakdown": {
            "tokens_per_sec": 4.23,
            "usd_per_hour": 0.1431,
            "usd_per_million": 9.3918,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 7.0894
              },
              {
                "resource": "Memory",
                "quantity": "11.2 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 9.326,
                "usd_per_million": 2.2039
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0985
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-15T06:44:44.996Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:0024080112206eef6b7384b664280f7a622b6c1a831a453029d0980956ec8090954e72662a4e",
            "signature": "PYnjyde4KfUVIYw8HXMH94/Msu3rxXo+eruh51LVfPGig4X+RjfXQH0jb65bpAEnsdI7hszcNSo41TisnozwDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "ywqajNnr4KJBorDevz6q9yIHFs6bw29WBmgtvn00KTk="
          },
          "build_page": "/inferences/gpt-oss/20b~davidau-openai-claude-4-5-opus-heretic-uncensored/mradermacher-iq1_m-ollama",
          "hub_url": "https://huggingface.co/mradermacher/openai-gpt-oss-20b-claude-4.5-opus-heretic-uncensored-i1-gguf",
          "weeks": [
            "2026-W46"
          ],
          "providers": [
            "662a4e"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/DavidAU/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "4.8 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "openai_gpt-oss-20b-GGUF-MXFP4-Experimental",
      "made_by": "bartowski",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68783facef79a05727260de3/UPX5RQxiPGA-ZbBmArIKq.png",
      "based_on": "gpt-oss 20B",
      "variant_of": "/inferences/gpt-oss/20b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/gpt-oss/20b~bartowski-openai-experimental",
      "hub_url": "https://huggingface.co/openai/gpt-oss-20b",
      "download_gb": 11.3,
      "params_b": 20,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 4.32,
        "context_tokens": null,
        "build": "MXFP4 (4-bit) by bartowski",
        "ref": "hf.co/bartowski/openai_gpt-oss-20b-GGUF-MXFP4-Experimental:MXFP4",
        "engine": "ollama-host",
        "cost_usd_per_million": 9.2234,
        "cost_breakdown": {
          "tokens_per_sec": 4.32,
          "usd_per_hour": 0.1433,
          "usd_per_million": 9.2234,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 6.9497
            },
            {
              "resource": "Memory",
              "quantity": "11.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 9.398,
              "usd_per_million": 2.1771
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0966
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T06:35:24.096Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "ONqpEZ/YbmMbLPLlvr6GytD+vdXFcuCBgeLje1HSOvKuHBTCxHqvcT+bXrzjxoQVKn3OREeZ9RJ+FNMA2bfUDQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "SXkhE2immoEOD7ZrlPMOMqIQe9WSF+He+zOFn6OEDcE="
        },
        "build_page": "/inferences/gpt-oss/20b~bartowski-openai-experimental/bartowski-mxfp4-ollama",
        "hub_url": "https://huggingface.co/bartowski/openai_gpt-oss-20b-gguf-mxfp4-experimental",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 4.32,
        "context_tokens": null,
        "build": "MXFP4 (4-bit) by bartowski",
        "ref": "hf.co/bartowski/openai_gpt-oss-20b-GGUF-MXFP4-Experimental:MXFP4",
        "engine": "ollama-host",
        "cost_usd_per_million": 9.2234,
        "cost_breakdown": {
          "tokens_per_sec": 4.32,
          "usd_per_hour": 0.1433,
          "usd_per_million": 9.2234,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 6.9497
            },
            {
              "resource": "Memory",
              "quantity": "11.3 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 9.398,
              "usd_per_million": 2.1771
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.0966
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T06:35:24.096Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "ONqpEZ/YbmMbLPLlvr6GytD+vdXFcuCBgeLje1HSOvKuHBTCxHqvcT+bXrzjxoQVKn3OREeZ9RJ+FNMA2bfUDQ==",
          "signing_payload_b64": "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",
          "payload_sha256": "SXkhE2immoEOD7ZrlPMOMqIQe9WSF+He+zOFn6OEDcE="
        },
        "build_page": "/inferences/gpt-oss/20b~bartowski-openai-experimental/bartowski-mxfp4-ollama",
        "hub_url": "https://huggingface.co/bartowski/openai_gpt-oss-20b-gguf-mxfp4-experimental",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 4.32,
          "context_tokens": null,
          "build": "MXFP4 (4-bit) by bartowski",
          "ref": "hf.co/bartowski/openai_gpt-oss-20b-GGUF-MXFP4-Experimental:MXFP4",
          "engine": "ollama-host",
          "cost_usd_per_million": 9.2234,
          "cost_breakdown": {
            "tokens_per_sec": 4.32,
            "usd_per_hour": 0.1433,
            "usd_per_million": 9.2234,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 6.9497
              },
              {
                "resource": "Memory",
                "quantity": "11.3 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 9.398,
                "usd_per_million": 2.1771
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.0966
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T06:35:24.096Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "ONqpEZ/YbmMbLPLlvr6GytD+vdXFcuCBgeLje1HSOvKuHBTCxHqvcT+bXrzjxoQVKn3OREeZ9RJ+FNMA2bfUDQ==",
            "signing_payload_b64": "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",
            "payload_sha256": "SXkhE2immoEOD7ZrlPMOMqIQe9WSF+He+zOFn6OEDcE="
          },
          "build_page": "/inferences/gpt-oss/20b~bartowski-openai-experimental/bartowski-mxfp4-ollama",
          "hub_url": "https://huggingface.co/bartowski/openai_gpt-oss-20b-gguf-mxfp4-experimental",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "openai/gpt-oss-20b",
        "url": "https://openrouter.ai/openai/gpt-oss-20b",
        "input_usd_per_million": 0.03,
        "output_usd_per_million": 0.13,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.02,
            "usd_per_m_out": 0.03,
            "context_length": 131072,
            "uptime_1d": 99.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.018,
            "usd_per_m_out": 0.09,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.018,
            "usd_per_m_out": 0.09,
            "context_length": 131072,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16/flex",
            "quantization": "bf16",
            "usd_per_m_in": 0.024,
            "usd_per_m_out": 0.112,
            "context_length": 131072,
            "uptime_1d": 91.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.13,
            "context_length": 131072,
            "uptime_1d": 93.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.029,
            "usd_per_m_out": 0.14,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.14,
            "context_length": 131072,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.15,
            "context_length": 131072,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock/eu-west-1",
            "quantization": null,
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.15,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock",
            "quantization": null,
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.15,
            "context_length": 131072,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 0.18,
            "context_length": 131072,
            "uptime_1d": 90.6,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Groq",
            "tag": "groq",
            "quantization": null,
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.3,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 12
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openai/gpt-oss-20b"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "4.3 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. About $9.22 per million tokens here, against OpenRouter's $0.13: the API is 71× cheaper. Run it here for privacy or to keep work on your own machines, not to save money."
    },
    {
      "model": "gpt-oss-20b-Derestricted",
      "made_by": "ArliAI",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6625f4a8a8d1362ebcc3851a/47WrP6UrAJPBboYMnA36j.png",
      "based_on": "gpt-oss 20B",
      "variant_of": "/inferences/gpt-oss/20b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/gpt-oss/20b~arliai-derestricted",
      "hub_url": "https://huggingface.co/ArliAI/gpt-oss-20b-Derestricted",
      "download_gb": 15.7,
      "params_b": 20,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 3.44,
        "context_tokens": null,
        "build": "Q5_K_M (5-bit) by MaziyarPanahi",
        "ref": "hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q5_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 12.6617,
        "cost_breakdown": {
          "tokens_per_sec": 3.44,
          "usd_per_hour": 0.1567,
          "usd_per_million": 12.6617,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 8.7268
            },
            {
              "resource": "Memory",
              "quantity": "15.7 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 13.11,
              "usd_per_million": 3.8136
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.1213
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T06:35:24.096Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "wjZhPCq87FwRwC3fOzwkRRxkttNBG+RuwgmnpOjmtmcCjlXlHeQW7GmZFAZFV7nL7Zl14KQp3KT7vGRzqSySAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "wWbY3d/Hjew+8RmG7ErFXuDBvttPkje/+j7czbXyeOc="
        },
        "build_page": "/inferences/gpt-oss/20b~arliai-derestricted/maziyarpanahi-q5_k_m-ollama",
        "hub_url": "https://huggingface.co/maziyarpanahi/gpt-oss-20b-derestricted-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 3.44,
        "context_tokens": null,
        "build": "Q5_K_M (5-bit) by MaziyarPanahi",
        "ref": "hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q5_K_M",
        "engine": "ollama-host",
        "cost_usd_per_million": 12.6617,
        "cost_breakdown": {
          "tokens_per_sec": 3.44,
          "usd_per_hour": 0.1567,
          "usd_per_million": 12.6617,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 8.7268
            },
            {
              "resource": "Memory",
              "quantity": "15.7 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 13.11,
              "usd_per_million": 3.8136
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.1213
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T06:35:24.096Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "wjZhPCq87FwRwC3fOzwkRRxkttNBG+RuwgmnpOjmtmcCjlXlHeQW7GmZFAZFV7nL7Zl14KQp3KT7vGRzqSySAw==",
          "signing_payload_b64": "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",
          "payload_sha256": "wWbY3d/Hjew+8RmG7ErFXuDBvttPkje/+j7czbXyeOc="
        },
        "build_page": "/inferences/gpt-oss/20b~arliai-derestricted/maziyarpanahi-q5_k_m-ollama",
        "hub_url": "https://huggingface.co/maziyarpanahi/gpt-oss-20b-derestricted-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 3.44,
          "context_tokens": null,
          "build": "Q5_K_M (5-bit) by MaziyarPanahi",
          "ref": "hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q5_K_M",
          "engine": "ollama-host",
          "cost_usd_per_million": 12.6617,
          "cost_breakdown": {
            "tokens_per_sec": 3.44,
            "usd_per_hour": 0.1567,
            "usd_per_million": 12.6617,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 8.7268
              },
              {
                "resource": "Memory",
                "quantity": "15.7 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 13.11,
                "usd_per_million": 3.8136
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1213
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T06:35:24.096Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "wjZhPCq87FwRwC3fOzwkRRxkttNBG+RuwgmnpOjmtmcCjlXlHeQW7GmZFAZFV7nL7Zl14KQp3KT7vGRzqSySAw==",
            "signing_payload_b64": "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",
            "payload_sha256": "wWbY3d/Hjew+8RmG7ErFXuDBvttPkje/+j7czbXyeOc="
          },
          "build_page": "/inferences/gpt-oss/20b~arliai-derestricted/maziyarpanahi-q5_k_m-ollama",
          "hub_url": "https://huggingface.co/maziyarpanahi/gpt-oss-20b-derestricted-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/ArliAI/gpt-oss-20b-Derestricted"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "3.4 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "gpt-oss-20b-heretic",
      "made_by": "p-e-w",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68ce0ae4ce1399b97e364af4/0sUhuSJaFvxPNSR8APc3V.png",
      "based_on": "gpt-oss 20B",
      "variant_of": "/inferences/gpt-oss/20b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/gpt-oss/20b~p-e-w-heretic",
      "hub_url": "https://huggingface.co/p-e-w/gpt-oss-20b-heretic",
      "download_gb": 20.6,
      "params_b": 20,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 3.36,
        "context_tokens": null,
        "build": "Q8_0 (8-bit) by DavidAU",
        "ref": "hf.co/DavidAU/OpenAi-GPT-oss-20b-HERETIC-uncensored-NEO-Imatrix-gguf:Q8_0",
        "engine": "ollama-host",
        "cost_usd_per_million": 14.1515,
        "cost_breakdown": {
          "tokens_per_sec": 3.36,
          "usd_per_hour": 0.1712,
          "usd_per_million": 14.1515,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 8.9276
            },
            {
              "resource": "Memory",
              "quantity": "20.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 17.137,
              "usd_per_million": 5.0997
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.1241
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T06:35:24.096Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "9CtCK+QPb2aORsG3/0sopimYsJBzsnrgxjjHiSMZwvT3ePHGAaQUA4d9JwfvVwWu0VeLXUvc+vRtFzu+P12qBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "GkcVC3nJaiPx/dAJIT/SlZlwhZ7LJ/mXI2FGqOej6jI="
        },
        "build_page": "/inferences/gpt-oss/20b~p-e-w-heretic/davidau-q8_0-ollama",
        "hub_url": "https://huggingface.co/davidau/openai-gpt-oss-20b-heretic-uncensored-neo-imatrix-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 3.36,
        "context_tokens": null,
        "build": "Q8_0 (8-bit) by DavidAU",
        "ref": "hf.co/DavidAU/OpenAi-GPT-oss-20b-HERETIC-uncensored-NEO-Imatrix-gguf:Q8_0",
        "engine": "ollama-host",
        "cost_usd_per_million": 14.1515,
        "cost_breakdown": {
          "tokens_per_sec": 3.36,
          "usd_per_hour": 0.1712,
          "usd_per_million": 14.1515,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 8.9276
            },
            {
              "resource": "Memory",
              "quantity": "20.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 17.137,
              "usd_per_million": 5.0997
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.1241
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T06:35:24.096Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "9CtCK+QPb2aORsG3/0sopimYsJBzsnrgxjjHiSMZwvT3ePHGAaQUA4d9JwfvVwWu0VeLXUvc+vRtFzu+P12qBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "GkcVC3nJaiPx/dAJIT/SlZlwhZ7LJ/mXI2FGqOej6jI="
        },
        "build_page": "/inferences/gpt-oss/20b~p-e-w-heretic/davidau-q8_0-ollama",
        "hub_url": "https://huggingface.co/davidau/openai-gpt-oss-20b-heretic-uncensored-neo-imatrix-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 3.36,
          "context_tokens": null,
          "build": "Q8_0 (8-bit) by DavidAU",
          "ref": "hf.co/DavidAU/OpenAi-GPT-oss-20b-HERETIC-uncensored-NEO-Imatrix-gguf:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 14.1515,
          "cost_breakdown": {
            "tokens_per_sec": 3.36,
            "usd_per_hour": 0.1712,
            "usd_per_million": 14.1515,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 8.9276
              },
              {
                "resource": "Memory",
                "quantity": "20.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 17.137,
                "usd_per_million": 5.0997
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1241
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T06:35:24.096Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "9CtCK+QPb2aORsG3/0sopimYsJBzsnrgxjjHiSMZwvT3ePHGAaQUA4d9JwfvVwWu0VeLXUvc+vRtFzu+P12qBg==",
            "signing_payload_b64": "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",
            "payload_sha256": "GkcVC3nJaiPx/dAJIT/SlZlwhZ7LJ/mXI2FGqOej6jI="
          },
          "build_page": "/inferences/gpt-oss/20b~p-e-w-heretic/davidau-q8_0-ollama",
          "hub_url": "https://huggingface.co/davidau/openai-gpt-oss-20b-heretic-uncensored-neo-imatrix-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/p-e-w/gpt-oss-20b-heretic"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "3.4 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Huihui-gpt-oss-20b-BF16-abliterated",
      "made_by": "huihui-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/66228dc4c8920ec3513dc81a/9stczijVqp6TgQduzT8kN.png",
      "based_on": "gpt-oss 20B",
      "variant_of": "/inferences/gpt-oss/20b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/gpt-oss/20b~huihui-ai-abliterated",
      "hub_url": "https://huggingface.co/huihui-ai/Huihui-gpt-oss-20b-BF16-abliterated",
      "download_gb": 20.6,
      "params_b": 20,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 3.28,
        "context_tokens": null,
        "build": "Q8_0 (8-bit) by DavidAU",
        "ref": "hf.co/DavidAU/OpenAi-GPT-oss-20b-abliterated-uncensored-NEO-Imatrix-gguf:Q8_0",
        "engine": "ollama-host",
        "cost_usd_per_million": 14.5051,
        "cost_breakdown": {
          "tokens_per_sec": 3.28,
          "usd_per_hour": 0.1712,
          "usd_per_million": 14.5051,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 9.1507
            },
            {
              "resource": "Memory",
              "quantity": "20.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 17.137,
              "usd_per_million": 5.2272
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.1272
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T06:35:24.096Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "7fq/nfUOIwWb1MEiE8f3ep48tfNxf1eW2P5+nkOEhpuA0tZQdKVuXMOOMHXxtCbew5+WpXV9nail1XqzYZArBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "f7DPvL366THj5jr8nPyomgXT9C+aDUS1tA6+H5oQaB8="
        },
        "build_page": "/inferences/gpt-oss/20b~huihui-ai-abliterated/davidau-q8_0-ollama",
        "hub_url": "https://huggingface.co/davidau/openai-gpt-oss-20b-abliterated-uncensored-neo-imatrix-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": "Intel Xeon",
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": "Kaggle",
          "vendor_org": null,
          "host_org": "kaggle",
          "key": "linux|amd64|4|31|none|0|||Kaggle"
        },
        "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
        "machines": 1,
        "tokens_per_sec": 3.28,
        "context_tokens": null,
        "build": "Q8_0 (8-bit) by DavidAU",
        "ref": "hf.co/DavidAU/OpenAi-GPT-oss-20b-abliterated-uncensored-NEO-Imatrix-gguf:Q8_0",
        "engine": "ollama-host",
        "cost_usd_per_million": 14.5051,
        "cost_breakdown": {
          "tokens_per_sec": 3.28,
          "usd_per_hour": 0.1712,
          "usd_per_million": 14.5051,
          "lines": [
            {
              "resource": "CPU",
              "quantity": "3 of 4 threads (the engine drives every thread but one)",
              "rate": "$0.036 per thread-hour",
              "uusd_per_second": 30,
              "usd_per_million": 9.1507
            },
            {
              "resource": "Memory",
              "quantity": "20.6 GB of weights held resident",
              "rate": "$0.003 per GB-hour",
              "uusd_per_second": 17.137,
              "usd_per_million": 5.2272
            },
            {
              "resource": "Energy",
              "quantity": "15 W modelled (3 threads × 5 W)",
              "rate": "$0.10 per kWh",
              "uusd_per_second": 0.417,
              "usd_per_million": 0.1272
            }
          ],
          "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
          "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
        },
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T06:35:24.096Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
          "signature": "7fq/nfUOIwWb1MEiE8f3ep48tfNxf1eW2P5+nkOEhpuA0tZQdKVuXMOOMHXxtCbew5+WpXV9nail1XqzYZArBg==",
          "signing_payload_b64": "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",
          "payload_sha256": "f7DPvL366THj5jr8nPyomgXT9C+aDUS1tA6+H5oQaB8="
        },
        "build_page": "/inferences/gpt-oss/20b~huihui-ai-abliterated/davidau-q8_0-ollama",
        "hub_url": "https://huggingface.co/davidau/openai-gpt-oss-20b-abliterated-uncensored-neo-imatrix-gguf",
        "weeks": [
          "2026-W44"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "Intel Xeon",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": 3.28,
          "context_tokens": null,
          "build": "Q8_0 (8-bit) by DavidAU",
          "ref": "hf.co/DavidAU/OpenAi-GPT-oss-20b-abliterated-uncensored-NEO-Imatrix-gguf:Q8_0",
          "engine": "ollama-host",
          "cost_usd_per_million": 14.5051,
          "cost_breakdown": {
            "tokens_per_sec": 3.28,
            "usd_per_hour": 0.1712,
            "usd_per_million": 14.5051,
            "lines": [
              {
                "resource": "CPU",
                "quantity": "3 of 4 threads (the engine drives every thread but one)",
                "rate": "$0.036 per thread-hour",
                "uusd_per_second": 30,
                "usd_per_million": 9.1507
              },
              {
                "resource": "Memory",
                "quantity": "20.6 GB of weights held resident",
                "rate": "$0.003 per GB-hour",
                "uusd_per_second": 17.137,
                "usd_per_million": 5.2272
              },
              {
                "resource": "Energy",
                "quantity": "15 W modelled (3 threads × 5 W)",
                "rate": "$0.10 per kWh",
                "uusd_per_second": 0.417,
                "usd_per_million": 0.1272
              }
            ],
            "formula": "USD per million tokens = metered µUSD per second ÷ tokens per second",
            "source": "epn-daemon config.DefaultPriceFloors and internal/cmd/token_price.go"
          },
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T06:35:24.096Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220a484f8a0a3e42726a38e1f1577f5bee530e51857184a81bae8967635d6e3654f",
            "signature": "7fq/nfUOIwWb1MEiE8f3ep48tfNxf1eW2P5+nkOEhpuA0tZQdKVuXMOOMHXxtCbew5+WpXV9nail1XqzYZArBg==",
            "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6My4yNzg0MzMxOTE2MTc4MywidW5pdCI6InRva2Vucy9zIiwidHMiOjE3OTMzNDIxMjQwOTYwMDQ3NTgsIm5vZGVfZGlkIjoiZGlkOmVwbjowMDI0MDgwMTEyMjBhNDg0ZjhhMGEzZTQyNzI2YTM4ZTFmMTU3N2Y1YmVlNTMwZTUxODU3MTg0YTgxYmFlODk2NzYzNWQ2ZTM2NTRmIiwiZXh0cmEiOnsiaGFyZHdhcmVfY2xhc3MiOiJsaW51eC9hbWQ2NC9ub25lLzMyZyIsIm1vZGVsIjoiaGYuY28vRGF2aWRBVS9PcGVuQWktR1BULW9zcy0yMGItYWJsaXRlcmF0ZWQtdW5jZW5zb3JlZC1ORU8tSW1hdHJpeC1nZ3VmOlE4XzAiLCJub3RlIjoic2lnbmVkIHJvbGxpbmcgb2JzZXJ2YXRpb24gZnJvbSByZWFsIGxvY2FsIGluZmVyZW5jZSBtZXRyaWNzOyBDTEkgYmVuY2ggaW5mZXJlbmNlIHJlbWFpbnMgU0EtQkVOQ0gtNSIsInJ1bnRpbWUiOiJvbGxhbWEtaG9zdCIsInNhbXBsZV9jb3VudCI6NTEsInNvdXJjZSI6Im1ldHJpY3MubW9kZWxzX3Rva2Vuc19wZXJfc2VjIiwidG90YWxfc2Vjb25kcyI6MTcyNC45MDkzMjk5OTkwMDE2LCJ0b3RhbF90b2tlbnMiOjU2NTV9fQ==",
            "payload_sha256": "f7DPvL366THj5jr8nPyomgXT9C+aDUS1tA6+H5oQaB8="
          },
          "build_page": "/inferences/gpt-oss/20b~huihui-ai-abliterated/davidau-q8_0-ollama",
          "hub_url": "https://huggingface.co/davidau/openai-gpt-oss-20b-abliterated-uncensored-neo-imatrix-gguf",
          "weeks": [
            "2026-W44"
          ],
          "providers": [
            "e3654f"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/huihui-ai/Huihui-gpt-oss-20b-BF16-abliterated"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "3.3 tokens a second on Linux · no GPU · 4 CPUs · 31 GB RAM. No hosted API sells it, so this network is the place to run it."
    },
    {
      "model": "Qwen 3.5 9B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3-5/9b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.5-9B",
      "download_gb": 5.3,
      "params_b": 9,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "machine not reported",
        "machine_detail": null,
        "machine_class": "unknown",
        "machines": 2,
        "tokens_per_sec": 2.39,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by unsloth",
        "ref": "hf.co/unsloth/Qwen3.5-9B-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-30T04:00:04.405Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "hW44MIAlhpIPtlh9xuVlT/Dpy0a0MTg5/IgLpjdl8QkdYc3DNhGfreArnIAUjTVTFDj119t4qsBiLeusr8oZDg==",
          "signing_payload_b64": "eyJtZXRyaWMiOiJpbmZlcmVuY2UudG9rZW5zX3Blcl9zZWMiLCJ2YWx1ZSI6Mi4zOTI0MjQyNzU3NzY1MjcsInVuaXQiOiJ0b2tlbnMvcyIsInRzIjoxNzkzMzMyODA0NDA1Mjc0Njg0LCJub2RlX2RpZCI6ImRpZDplcG46MDAyNDA4MDExMjIwZWUzNWRjNDZhMjIxMmNiMzE0NDRjYTRjMTk2NjlhYzhkOTdiMDkyNTIwYjhjOWU0ZDI5YWI0OGQ5ZWU2MWQzYiIsImV4dHJhIjp7ImhhcmR3YXJlX2NsYXNzIjoibGludXgvYW1kNjQvbm9uZS8zMmciLCJtb2RlbCI6ImhmLmNvL3Vuc2xvdGgvUXdlbjMuNS05Qi1HR1VGOlE0X0tfTSIsIm5vdGUiOiJzaWduZWQgcm9sbGluZyBvYnNlcnZhdGlvbiBmcm9tIHJlYWwgbG9jYWwgaW5mZXJlbmNlIG1ldHJpY3M7IENMSSBiZW5jaCBpbmZlcmVuY2UgcmVtYWlucyBTQS1CRU5DSC01IiwicnVudGltZSI6Im9sbGFtYS1ob3N0Iiwic2FtcGxlX2NvdW50IjoxMCwic291cmNlIjoibWV0cmljcy5tb2RlbHNfdG9rZW5zX3Blcl9zZWMiLCJ0b3RhbF9zZWNvbmRzIjoyMzU5LjExMzMyOTk5OTk5OTcsInRvdGFsX3Rva2VucyI6NTY0NH19",
          "payload_sha256": "oz8IG1hK8gawyLDYnAqOzN4JM0bBpyglTJi6n/FlKFc="
        },
        "build_page": "/inferences/qwen-3-5/9b/unsloth-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/unsloth/qwen3.5-9b-gguf",
        "weeks": [],
        "providers": [
          "e61d3b",
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 2,
      "weeks_measured": [],
      "measured_on": [
        {
          "machine": "machine not reported",
          "machine_detail": null,
          "machine_class": "unknown",
          "machines": 2,
          "tokens_per_sec": 2.39,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by unsloth",
          "ref": "hf.co/unsloth/Qwen3.5-9B-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-30T04:00:04.405Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "hW44MIAlhpIPtlh9xuVlT/Dpy0a0MTg5/IgLpjdl8QkdYc3DNhGfreArnIAUjTVTFDj119t4qsBiLeusr8oZDg==",
            "signing_payload_b64": "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",
            "payload_sha256": "oz8IG1hK8gawyLDYnAqOzN4JM0bBpyglTJi6n/FlKFc="
          },
          "build_page": "/inferences/qwen-3-5/9b/unsloth-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/unsloth/qwen3.5-9b-gguf",
          "weeks": [],
          "providers": [
            "e61d3b",
            "e3654f"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.5-9b",
        "url": "https://openrouter.ai/qwen/qwen3.5-9b",
        "input_usd_per_million": 0.037,
        "output_usd_per_million": 0.054,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16/flex",
            "quantization": "bf16",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.12,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.13,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.15,
            "context_length": 262144,
            "uptime_1d": 57.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.15,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.148,
            "usd_per_m_out": 0.218,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16/priority",
            "quantization": "bf16",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.225,
            "context_length": 262144,
            "uptime_1d": 96.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.25,
            "context_length": 262144,
            "uptime_1d": 60.4,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.17,
            "usd_per_m_out": 0.25,
            "context_length": 262144,
            "uptime_1d": 97.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 8
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.5-9B"
      },
      "builds": 3,
      "variants": [
        {
          "model": "Qwen3.5-9B-Artemis-GGUF",
          "made_by": "turintech",
          "page": "/inferences/qwen-3-5/9b~turintech-artemis",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "2.4 tokens a second on a machine not reported. OpenRouter sells it at $0.05 per million output tokens."
    },
    {
      "model": "Granite 4.2 8B",
      "made_by": "ibm-granite",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/639bcaa2445b133a4e942436/CEW-OjXkRkDNmTxSu8Egh.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/granite-4-2/8b",
      "hub_url": "https://huggingface.co/ibm-granite/granite-4.2-8b",
      "download_gb": 4.6,
      "params_b": 8,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "machine not reported",
        "machine_detail": null,
        "machine_class": "unknown",
        "machines": 1,
        "tokens_per_sec": 1.87,
        "context_tokens": null,
        "build": "Q4_K_M (4-bit) by ibm-granite",
        "ref": "hf.co/ibm-granite/granite-4.2-8b-GGUF:Q4_K_M",
        "engine": "ollama",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": null,
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-31T00:18:37.298Z",
        "signed": true,
        "proof": {
          "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
          "signature": "icyyvBZSBH8lH1/TgpSefrdPBjZGm7uw9HoF1d9lnZDlW29iDNxxyYJRj+QTBCuiLdOWLhXhTATNvxPWyvHjDw==",
          "signing_payload_b64": "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",
          "payload_sha256": "tcKInU+NfUeuEfNy92D6azixWacAx5MnDObyFi0Avdo="
        },
        "build_page": "/inferences/granite-4-2/8b/ibm-granite-q4_k_m-ollama",
        "hub_url": "https://huggingface.co/ibm-granite/granite-4.2-8b-gguf",
        "weeks": [],
        "providers": [
          "e61d3b"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [],
      "measured_on": [
        {
          "machine": "machine not reported",
          "machine_detail": null,
          "machine_class": "unknown",
          "machines": 1,
          "tokens_per_sec": 1.87,
          "context_tokens": null,
          "build": "Q4_K_M (4-bit) by ibm-granite",
          "ref": "hf.co/ibm-granite/granite-4.2-8b-GGUF:Q4_K_M",
          "engine": "ollama",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": null,
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-31T00:18:37.298Z",
          "signed": true,
          "proof": {
            "node_did": "did:epn:002408011220ee35dc46a2212cb31444ca4c19669ac8d97b092520b8c9e4d29ab48d9ee61d3b",
            "signature": "icyyvBZSBH8lH1/TgpSefrdPBjZGm7uw9HoF1d9lnZDlW29iDNxxyYJRj+QTBCuiLdOWLhXhTATNvxPWyvHjDw==",
            "signing_payload_b64": "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",
            "payload_sha256": "tcKInU+NfUeuEfNy92D6azixWacAx5MnDObyFi0Avdo="
          },
          "build_page": "/inferences/granite-4-2/8b/ibm-granite-q4_k_m-ollama",
          "hub_url": "https://huggingface.co/ibm-granite/granite-4.2-8b-gguf",
          "weeks": [],
          "providers": [
            "e61d3b"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "ibm-granite/granite-4.2-8b",
        "url": "https://openrouter.ai/ibm-granite/granite-4.2-8b",
        "input_usd_per_million": 0.06,
        "output_usd_per_million": 0.15,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "CoreWeave",
            "tag": "coreweave/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.15,
            "context_length": 131072,
            "uptime_1d": 93,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.25,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/ibm-granite/granite-4.2-8b"
      },
      "builds": 4,
      "variants": [],
      "bottom_line": "1.9 tokens a second on a machine not reported. OpenRouter sells it at $0.15 per million output tokens."
    },
    {
      "model": "5b",
      "made_by": "THUDM",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Image generation",
      "community_build": false,
      "page": "/inferences/cogvideox/5b",
      "hub_url": "https://huggingface.co/zai-org/CogVideoX-5b",
      "download_gb": 20,
      "params_b": 5,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "by zai-org",
        "ref": "zai-org/CogVideoX-5b",
        "engine": "diffusers",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-23T20:12:21.561Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/cogvideox/5b/zai-org-cogvideox-5b-diffusers",
        "hub_url": "https://huggingface.co/zai-org/cogvideox-5b",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by zai-org",
          "ref": "zai-org/CogVideoX-5b",
          "engine": "diffusers",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T20:12:21.561Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/cogvideox/5b/zai-org-cogvideox-5b-diffusers",
          "hub_url": "https://huggingface.co/zai-org/cogvideox-5b",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/zai-org/CogVideoX-5b"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "CogVideoX 2B",
      "made_by": "THUDM",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/63033dc4e1e7f0e03a5e1a31/0BibmRdezvN6v6d2CEtd5.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Image generation",
      "community_build": false,
      "page": "/inferences/cogvideox/2b",
      "hub_url": "https://huggingface.co/THUDM/CogVideoX-2b",
      "download_gb": 12.8,
      "params_b": 2,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 12,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|||"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "BF16 (16-bit, full precision) by THUDM",
        "ref": "THUDM/CogVideoX-2b",
        "engine": "diffusers",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-09-15T13:03:41.444Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/cogvideox/2b/thudm-bf16-diffusers",
        "hub_url": "https://huggingface.co/thudm/cogvideox-2b",
        "weeks": [
          "2026-W38",
          "2026-W43",
          "2026-W45"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "cheapest": null,
      "machine_kinds": 4,
      "machines_total": 5,
      "weeks_measured": [
        "2026-W38",
        "2026-W43",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "BF16 (16-bit, full precision) by THUDM",
          "ref": "THUDM/CogVideoX-2b",
          "engine": "diffusers",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-15T13:03:41.444Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/cogvideox/2b/thudm-bf16-diffusers",
          "hub_url": "https://huggingface.co/thudm/cogvideox-2b",
          "weeks": [
            "2026-W38",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "BF16 (16-bit, full precision) by THUDM",
          "ref": "THUDM/CogVideoX-2b",
          "engine": "diffusers",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-17T01:06:35.345Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/cogvideox/2b/thudm-bf16-diffusers",
          "hub_url": "https://huggingface.co/thudm/cogvideox-2b",
          "weeks": [
            "2026-W38",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "BF16 (16-bit, full precision) by THUDM",
          "ref": "THUDM/CogVideoX-2b",
          "engine": "diffusers",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-21T06:35:45.949Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/cogvideox/2b/thudm-bf16-diffusers",
          "hub_url": "https://huggingface.co/thudm/cogvideox-2b",
          "weeks": [
            "2026-W38",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "BF16 (16-bit, full precision) by THUDM",
          "ref": "THUDM/CogVideoX-2b",
          "engine": "diffusers",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T12:09:40.542Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/cogvideox/2b/thudm-bf16-diffusers",
          "hub_url": "https://huggingface.co/thudm/cogvideox-2b",
          "weeks": [
            "2026-W38",
            "2026-W43",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "by zai-org",
          "ref": "zai-org/CogVideoX-2b",
          "engine": "diffusers",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-21T06:34:26.383Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/cogvideox/2b/zai-org-cogvideox-2b-diffusers",
          "hub_url": "https://huggingface.co/zai-org/cogvideox-2b",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/THUDM/CogVideoX-2b"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "Runs on 4 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Depth Anything V2 Base",
      "made_by": "depth-anything",
      "maker_logo": "https://www.gravatar.com/avatar/f4f72407f9653fd54129eee82ed335fa?d=retro&size=100",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Vision (counting, locating)",
      "community_build": false,
      "page": "/inferences/depth-anything/v2-base",
      "hub_url": "https://huggingface.co/depth-anything/Depth-Anything-V2-Base-hf",
      "download_gb": 0.4,
      "params_b": 0.1,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 2,
          "host": null,
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|2|||"
        },
        "machine_class": "windows|amd64|8|16|intel|2|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by depth-anything",
        "ref": "depth-anything/Depth-Anything-V2-Base-hf",
        "engine": "vision-task",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T11:18:15.444Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/depth-anything/v2-base/depth-anything-fp32-vision-task",
        "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-base-hf",
        "weeks": [
          "2026-W36",
          "2026-W43",
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 8,
      "machines_total": 8,
      "weeks_measured": [
        "2026-W36",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Base-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T11:18:15.444Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-base/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-base-hf",
          "weeks": [
            "2026-W36",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Base-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-22T08:18:20.813Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-base/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-base-hf",
          "weeks": [
            "2026-W36",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Base-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T11:40:20.408Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-base/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-base-hf",
          "weeks": [
            "2026-W36",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.22631",
            "cpus": 8,
            "cpu_model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
            "ram_gb": 7,
            "gpu_vendor": "AMD",
            "gpu_model": "AMD Radeon(TM) Vega 8 Mobile Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "amd",
            "host_org": null,
            "key": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Base-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 997.6,
              "multi_core_hashes_per_sec": 8516.9,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T08:14:32.559Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "memory": {
              "total_gb": 7,
              "bandwidth_gb_s": 6,
              "latency_ns": 61.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T08:14:33.251Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2347.3,
              "write_mb_s": 204.2,
              "random_read_iops": 96070,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-10-29T08:14:33.869Z",
              "measured_in": "IN_560001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-28T13:22:06.050Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-base/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-base-hf",
          "weeks": [
            "2026-W36",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "d061c9"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Base-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T02:43:11.525Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-base/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-base-hf",
          "weeks": [
            "2026-W36",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "AMD EPYC 7B12",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Base-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T21:38:52.035Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-base/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-base-hf",
          "weeks": [
            "2026-W36",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e"
          ]
        },
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Base-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-08T19:04:06.193Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-base/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-base-hf",
          "weeks": [
            "2026-W36",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Base-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T13:16:15.933Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-base/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-base-hf",
          "weeks": [
            "2026-W36",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/depth-anything/Depth-Anything-V2-Base-hf"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 8 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Depth Anything V2 Small",
      "made_by": "depth-anything",
      "maker_logo": "https://www.gravatar.com/avatar/f4f72407f9653fd54129eee82ed335fa?d=retro&size=100",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Vision (counting, locating)",
      "community_build": false,
      "page": "/inferences/depth-anything/v2-small",
      "hub_url": "https://huggingface.co/depth-anything/Depth-Anything-V2-Small-hf",
      "download_gb": 0.1,
      "params_b": 0.03,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 2,
          "host": null,
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|2|||"
        },
        "machine_class": "windows|amd64|8|16|intel|2|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by depth-anything",
        "ref": "depth-anything/Depth-Anything-V2-Small-hf",
        "engine": "vision-task",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T17:30:35.747Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/depth-anything/v2-small/depth-anything-fp32-vision-task",
        "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-small-hf",
        "weeks": [
          "2026-W36",
          "2026-W37",
          "2026-W38",
          "2026-W43",
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 8,
      "machines_total": 8,
      "weeks_measured": [
        "2026-W36",
        "2026-W37",
        "2026-W38",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Small-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T17:30:35.747Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-small/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-small-hf",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Small-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-08T19:07:48.995Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-small/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-small-hf",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Small-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-21T06:57:59.068Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-small/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-small-hf",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "a5be46"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Small-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 181,
              "multi_core_hashes_per_sec": 436.8,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-30T20:43:55.306Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 3.4,
              "latency_ns": 49.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-30T20:43:56.203Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4271.7,
              "write_mb_s": 177.9,
              "random_read_iops": 391634,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T20:43:56.799Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T11:40:54.203Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-small/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-small-hf",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e61d3b"
          ]
        },
        {
          "machine": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
          "machine_detail": {
            "name": "Windows · AMD Radeon(TM) Vega 8 Mobile Graphics 1 GB · 8 CPUs · 7 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.22631",
            "cpus": 8,
            "cpu_model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
            "ram_gb": 7,
            "gpu_vendor": "AMD",
            "gpu_model": "AMD Radeon(TM) Vega 8 Mobile Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "amd",
            "host_org": null,
            "key": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|7|amd|1|AMD Radeon(TM) Vega 8 Mobile Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Small-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "model_raw": "AMD Ryzen 5 2500U with Radeon Vega Mobile Gfx",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 997.6,
              "multi_core_hashes_per_sec": 8516.9,
              "sustained_ratio": 1.2,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T08:14:32.559Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "memory": {
              "total_gb": 7,
              "bandwidth_gb_s": 6,
              "latency_ns": 61.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T08:14:33.251Z",
              "measured_in": "IN_560001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2347.3,
              "write_mb_s": 204.2,
              "random_read_iops": 96070,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-10-29T08:14:33.869Z",
              "measured_in": "IN_560001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-28T13:24:41.440Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-small/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-small-hf",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "d061c9"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Small-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T02:43:30.231Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-small/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-small-hf",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "AMD EPYC 7B12",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Small-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T22:40:30.949Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-small/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-small-hf",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by depth-anything",
          "ref": "depth-anything/Depth-Anything-V2-Small-hf",
          "engine": "vision-task",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T13:18:34.441Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/depth-anything/v2-small/depth-anything-fp32-vision-task",
          "hub_url": "https://huggingface.co/depth-anything/depth-anything-v2-small-hf",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/depth-anything/Depth-Anything-V2-Small-hf"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 8 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "GLM-4.5-Air-0.72B-MTP",
      "made_by": "nm-testing",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/noauth/qQp5Mx0lVZQLcA4ZKPi40.png",
      "based_on": "GLM-4.5",
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-4-5/0.72b~nm-testing-air-mtp",
      "hub_url": "https://huggingface.co/nm-testing/GLM-4.5-Air-0.72B-MTP",
      "download_gb": 1.3,
      "params_b": 0.72,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
        "machine_detail": {
          "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 12,
          "cpu_model": "Intel Xeon",
          "ram_gb": 53,
          "gpu_vendor": "NVIDIA",
          "gpu_model": "L4",
          "gpu_count": 1,
          "vram_gb": 23,
          "host": "Google Colab",
          "vendor_org": "nvidia",
          "host_org": "google",
          "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
        },
        "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "BF16 (16-bit, full precision) by nm-testing",
        "ref": "nm-testing/GLM-4.5-Air-0.72B-MTP",
        "engine": "vllm",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 12,
            "cores_physical": 6,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 295.2,
            "multi_core_hashes_per_sec": 1700.6,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-07T04:46:20.920Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "memory": {
            "total_gb": 53,
            "bandwidth_gb_s": 2.8,
            "latency_ns": 23.6,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-07T04:46:21.530Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA L4",
            "memory_gb": 22,
            "fp16_tflops": 52.7,
            "watts": 34.2,
            "tflops_per_watt": 1.542,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-07T04:46:37.158Z",
            "measured_in": "SG_190464",
            "signed": true
          },
          "storage": {
            "read_mb_s": 3356.3,
            "write_mb_s": 287.7,
            "random_read_iops": 812404,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-07T04:46:22.036Z",
            "measured_in": "SG_190464",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-11-06T07:33:08.576Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/glm-4-5/0.72b~nm-testing-air-mtp/nm-testing-bf16-vllm",
        "hub_url": "https://huggingface.co/nm-testing/glm-4.5-air-0.72b-mtp",
        "weeks": [
          "2026-W45"
        ],
        "providers": [
          "dae679"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "BF16 (16-bit, full precision) by nm-testing",
          "ref": "nm-testing/GLM-4.5-Air-0.72B-MTP",
          "engine": "vllm",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T07:33:08.576Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/glm-4-5/0.72b~nm-testing-air-mtp/nm-testing-bf16-vllm",
          "hub_url": "https://huggingface.co/nm-testing/glm-4.5-air-0.72b-mtp",
          "weeks": [
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/nm-testing/GLM-4.5-Air-0.72B-MTP"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Ran on 1 kind of machine here, but its speed was not recorded."
    },
    {
      "model": "HunyuanVideo T2V 720p",
      "made_by": "tencent",
      "maker_logo": "https://www.gravatar.com/avatar/c98615f99e27d862af2babbd9fbaa72c?d=retro&size=100",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Image generation",
      "community_build": false,
      "page": "/inferences/hunyuan-video/t2v-720p",
      "hub_url": "https://huggingface.co/hunyuanvideo-community/HunyuanVideo",
      "download_gb": 39,
      "params_b": 13,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "BF16 (16-bit, full precision) by hunyuanvideo-community",
        "ref": "hunyuanvideo-community/HunyuanVideo",
        "engine": "diffusers",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-23T20:01:00.857Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/hunyuan-video/t2v-720p/hunyuanvideo-community-bf16-diffusers",
        "hub_url": "https://huggingface.co/hunyuanvideo-community/hunyuanvideo",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "BF16 (16-bit, full precision) by hunyuanvideo-community",
          "ref": "hunyuanvideo-community/HunyuanVideo",
          "engine": "diffusers",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-23T20:01:00.857Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/hunyuan-video/t2v-720p/hunyuanvideo-community-bf16-diffusers",
          "hub_url": "https://huggingface.co/hunyuanvideo-community/hunyuanvideo",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": "https://huggingface.co/hunyuanvideo-community/HunyuanVideo"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "LTX-Video 2B 0.9.5",
      "made_by": "Lightricks",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/669524bcbcd81f395e8f60f6/0ynfqKEWMh_dn3h4ff1K5.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Image generation",
      "community_build": false,
      "page": "/inferences/ltx-video/2b-0-9-5",
      "hub_url": "https://huggingface.co/Lightricks/LTX-Video-0.9.5",
      "download_gb": 23.6,
      "params_b": 2,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "BF16 (16-bit, full precision) by Lightricks",
        "ref": "Lightricks/LTX-Video-0.9.5",
        "engine": "diffusers",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-21T06:38:05.516Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/ltx-video/2b-0-9-5/lightricks-bf16-diffusers",
        "hub_url": "https://huggingface.co/lightricks/ltx-video-0.9.5",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "BF16 (16-bit, full precision) by Lightricks",
          "ref": "Lightricks/LTX-Video-0.9.5",
          "engine": "diffusers",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-21T06:38:05.516Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/ltx-video/2b-0-9-5/lightricks-bf16-diffusers",
          "hub_url": "https://huggingface.co/lightricks/ltx-video-0.9.5",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/Lightricks/LTX-Video-0.9.5"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "LTX-Video-ICLoRA-detailer-13b-0.9.8",
      "made_by": "lightricks",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/669524bcbcd81f395e8f60f6/0ynfqKEWMh_dn3h4ff1K5.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Other",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/Lightricks/LTX-Video-ICLoRA-detailer-13b-0.9.8",
      "download_gb": null,
      "params_b": 13,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "windows|amd64|8|16|none|0|||"
        },
        "machine_class": "windows|amd64|8|16|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "Lightricks/LTX-Video-ICLoRA-detailer-13b-0.9.8",
        "engine": null,
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-09-05T11:11:36.950Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/ltx-video/13b/lightricks-ltx-video-iclora-detailer-13b-0.9.8-unknown",
        "hub_url": "https://huggingface.co/lightricks/ltx-video-iclora-detailer-13b-0.9.8",
        "weeks": [
          "2026-W36"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W36"
      ],
      "measured_on": [
        {
          "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "windows|amd64|8|16|none|0|||"
          },
          "machine_class": "windows|amd64|8|16|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "Lightricks/LTX-Video-ICLoRA-detailer-13b-0.9.8",
          "engine": null,
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-05T11:11:36.950Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/ltx-video/13b/lightricks-ltx-video-iclora-detailer-13b-0.9.8-unknown",
          "hub_url": "https://huggingface.co/lightricks/ltx-video-iclora-detailer-13b-0.9.8",
          "weeks": [
            "2026-W36"
          ],
          "providers": [
            "9e3c79"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Madlad-400 3B MT",
      "made_by": "google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Translation",
      "community_build": false,
      "page": "/inferences/madlad/3b",
      "hub_url": "https://huggingface.co/google/madlad400-3b-mt",
      "download_gb": 11,
      "params_b": 3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [
        "Assamese",
        "Bengali",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Odia",
        "Punjabi",
        "Tamil"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by google",
        "ref": "google/madlad400-3b-mt",
        "engine": "translate-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T13:24:56.870Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/madlad/3b/google-fp32-translate-hf",
        "hub_url": "https://huggingface.co/google/madlad400-3b-mt",
        "weeks": [
          "2026-W43",
          "2026-W44",
          "2026-W45"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 2,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W43",
        "2026-W44",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by google",
          "ref": "google/madlad400-3b-mt",
          "engine": "translate-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T13:24:56.870Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/madlad/3b/google-fp32-translate-hf",
          "hub_url": "https://huggingface.co/google/madlad400-3b-mt",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "e3654f"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by google",
          "ref": "google/madlad400-3b-mt",
          "engine": "translate-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T18:10:50.632Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/madlad/3b/google-fp32-translate-hf",
          "hub_url": "https://huggingface.co/google/madlad400-3b-mt",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/google/madlad400-3b-mt"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 2 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Nllb-200 distilled 1.3B",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Translation",
      "community_build": false,
      "page": "/inferences/nllb/1.3b-distilled",
      "hub_url": "https://huggingface.co/facebook/nllb-200-distilled-1.3B",
      "download_gb": 5.1,
      "params_b": 1.3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [
        "Assamese",
        "Bengali",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Odia",
        "Punjabi",
        "Tamil",
        "Telugu",
        "Urdu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/nllb-200-distilled-1.3B",
        "engine": "translate-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:48:56.710Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/nllb/1.3b-distilled/facebook-fp32-translate-hf",
        "hub_url": "https://huggingface.co/facebook/nllb-200-distilled-1.3b",
        "weeks": [
          "2026-W43",
          "2026-W44",
          "2026-W45"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 3,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W43",
        "2026-W44",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/nllb-200-distilled-1.3B",
          "engine": "translate-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:48:56.710Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/nllb/1.3b-distilled/facebook-fp32-translate-hf",
          "hub_url": "https://huggingface.co/facebook/nllb-200-distilled-1.3b",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "e3654f"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/nllb-200-distilled-1.3B",
          "engine": "translate-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T02:49:47.929Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/nllb/1.3b-distilled/facebook-fp32-translate-hf",
          "hub_url": "https://huggingface.co/facebook/nllb-200-distilled-1.3b",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/nllb-200-distilled-1.3B",
          "engine": "translate-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-08T19:42:05.140Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/nllb/1.3b-distilled/facebook-fp32-translate-hf",
          "hub_url": "https://huggingface.co/facebook/nllb-200-distilled-1.3b",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "84bf54"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/nllb-200-distilled-1.3B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 3 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Nllb-200 distilled 600M",
      "made_by": "facebook",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1592839207516-noauth.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Translation",
      "community_build": false,
      "page": "/inferences/nllb/600m-distilled",
      "hub_url": "https://huggingface.co/facebook/nllb-200-distilled-600M",
      "download_gb": 2.3,
      "params_b": 0.6,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [
        "Assamese",
        "Bengali",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Odia",
        "Punjabi",
        "Tamil",
        "Telugu",
        "Urdu"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 2,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "FP32 (32-bit, full precision) by facebook",
        "ref": "facebook/nllb-200-distilled-600M",
        "engine": "translate-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T12:30:35.719Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/nllb/600m-distilled/facebook-fp32-translate-hf",
        "hub_url": "https://huggingface.co/facebook/nllb-200-distilled-600m",
        "weeks": [
          "2026-W43",
          "2026-W44",
          "2026-W45",
          "2026-W46"
        ],
        "providers": [
          "e3654f",
          "e61d3b"
        ]
      },
      "cheapest": null,
      "machine_kinds": 6,
      "machines_total": 8,
      "weeks_measured": [
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 2,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/nllb-200-distilled-600M",
          "engine": "translate-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T12:30:35.719Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/nllb/600m-distilled/facebook-fp32-translate-hf",
          "hub_url": "https://huggingface.co/facebook/nllb-200-distilled-600m",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "e3654f",
            "e61d3b"
          ]
        },
        {
          "machine": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) UHD Graphics 730 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": "12th Gen Intel Core i3-12100",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) UHD Graphics 730",
            "gpu_count": 1,
            "vram_gb": 2,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|2|Intel(R) UHD Graphics 730|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/nllb-200-distilled-600M",
          "engine": "translate-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-01T09:15:53.598Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/nllb/600m-distilled/facebook-fp32-translate-hf",
          "hub_url": "https://huggingface.co/facebook/nllb-200-distilled-600m",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/nllb-200-distilled-600M",
          "engine": "translate-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T02:46:04.134Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/nllb/600m-distilled/facebook-fp32-translate-hf",
          "hub_url": "https://huggingface.co/facebook/nllb-200-distilled-600m",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "dae679"
          ]
        },
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": "AMD EPYC 7B12",
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": "Kaggle",
            "vendor_org": null,
            "host_org": "kaggle",
            "key": "linux|amd64|4|31|none|0|||Kaggle"
          },
          "machine_class": "linux|amd64|4|31|none|0|||Kaggle",
          "machines": 2,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/nllb-200-distilled-600M",
          "engine": "translate-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD EPYC 7B12",
              "model_raw": "AMD EPYC 7B12",
              "clock_ghz": null,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 512,
              "l3_mb": 16,
              "single_core_hashes_per_sec": 1460.2,
              "multi_core_hashes_per_sec": 3246.1,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T01:38:20.212Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 6.3,
              "latency_ns": 21.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T01:38:21.135Z",
              "measured_in": "TW_110",
              "signed": true
            },
            "storage": {
              "read_mb_s": 4982.5,
              "write_mb_s": 179.3,
              "random_read_iops": 476287,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-15T01:38:21.741Z",
              "measured_in": "TW_110",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-14T21:50:20.448Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/nllb/600m-distilled/facebook-fp32-translate-hf",
          "hub_url": "https://huggingface.co/facebook/nllb-200-distilled-600m",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "662a4e",
            "4e94b2"
          ]
        },
        {
          "machine": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel(R) Iris(R) Xe Graphics 1 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26300",
            "cpus": 8,
            "cpu_model": "11th Gen Intel Core i5-1135G7",
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": "Intel(R) Iris(R) Xe Graphics",
            "gpu_count": 1,
            "vram_gb": 1,
            "host": "own machine",
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine"
          },
          "machine_class": "windows|amd64|8|16|intel|1|Intel(R) Iris(R) Xe Graphics|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/nllb-200-distilled-600M",
          "engine": "translate-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "11th Gen Intel Core i5-1135G7",
              "model_raw": "11th Gen Intel Core i5-1135G7",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 502.2,
              "multi_core_hashes_per_sec": 1190.3,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T12:56:42.595Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "memory": {
              "total_gb": 15.8,
              "bandwidth_gb_s": 1.2,
              "latency_ns": 253.1,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T12:56:43.806Z",
              "measured_in": "IN_500001",
              "signed": true
            },
            "storage": {
              "read_mb_s": 329.7,
              "write_mb_s": 12.4,
              "random_read_iops": 8193,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T12:56:49.437Z",
              "measured_in": "IN_500001",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-07T20:26:37.160Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/nllb/600m-distilled/facebook-fp32-translate-hf",
          "hub_url": "https://huggingface.co/facebook/nllb-200-distilled-600m",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "2e5c02"
          ]
        },
        {
          "machine": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · GeForce RTX 3080 Ti 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": "AMD Ryzen 7 5800X 8-Core",
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "GeForce RTX 3080 Ti",
            "gpu_count": 1,
            "vram_gb": 12,
            "host": "own machine",
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|GeForce RTX 3080 Ti|1|own machine",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "FP32 (32-bit, full precision) by facebook",
          "ref": "facebook/nllb-200-distilled-600M",
          "engine": "translate-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-10T09:40:00.877Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/nllb/600m-distilled/facebook-fp32-translate-hf",
          "hub_url": "https://huggingface.co/facebook/nllb-200-distilled-600m",
          "weeks": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "providers": [
            "84bf54"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": "CC BY-NC 4.0 (non-commercial)",
        "you_may_serve": "Only your own machines, not for paid work",
        "class": "non-commercial",
        "source": "https://huggingface.co/facebook/nllb-200-distilled-600M"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 6 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Pruna-Qwen-Image-2.1",
      "made_by": "prunaai",
      "maker_logo": null,
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Other",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/PrunaAI/Pruna-Qwen-Image-2.1",
      "download_gb": null,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "os": "Mac",
          "os_version": "macOS 26.5.2",
          "cpus": 10,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "Apple",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": null,
          "host": null,
          "vendor_org": "apple",
          "host_org": null,
          "key": "darwin|arm64|10|16|apple|0|||"
        },
        "machine_class": "darwin|arm64|10|16|apple|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "PrunaAI/Pruna-Qwen-Image-2.1",
        "engine": "generation",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Apple M4",
            "model_raw": "Apple M4",
            "clock_ghz": null,
            "cores_logical": 10,
            "cores_physical": 10,
            "threads_per_core": 1,
            "sockets": 1,
            "l2_kb": 4096,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2850.2,
            "multi_core_hashes_per_sec": 22838.7,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-02T08:29:50.654Z",
            "measured_in": "IN_110054",
            "signed": true
          },
          "memory": {
            "total_gb": 16,
            "bandwidth_gb_s": 42.7,
            "latency_ns": 6.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-02T08:29:51.180Z",
            "measured_in": "IN_110054",
            "signed": true
          },
          "storage": {
            "read_mb_s": 13543.1,
            "write_mb_s": 1696,
            "random_read_iops": 1733096,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-02T08:29:51.427Z",
            "measured_in": "IN_110054",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-16T07:34:24.723Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/qwen-image/pruna-2-1/prunaai-pruna-qwen-image-2.1-generation",
        "hub_url": "https://huggingface.co/prunaai/pruna-qwen-image-2.1",
        "weeks": [
          "2026-W40",
          "2026-W42"
        ],
        "providers": [
          "afbb8a"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W40",
        "2026-W42"
      ],
      "measured_on": [
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "PrunaAI/Pruna-Qwen-Image-2.1",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-16T07:34:24.723Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-image/pruna-2-1/prunaai-pruna-qwen-image-2.1-generation",
          "hub_url": "https://huggingface.co/prunaai/pruna-qwen-image-2.1",
          "weeks": [
            "2026-W40",
            "2026-W42"
          ],
          "providers": [
            "afbb8a"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "QVGen-CogVideoX-2B-W4A4",
      "made_by": "harahan",
      "maker_logo": null,
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Other",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/Harahan/QVGen-CogVideoX-2B-W4A4",
      "download_gb": null,
      "params_b": 2,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "windows|amd64|8|16|none|0|||"
        },
        "machine_class": "windows|amd64|8|16|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "Harahan/QVGen-CogVideoX-2B-W4A4",
        "engine": null,
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-09-05T11:13:13.736Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/cogvideox/2b/harahan-qvgen-cogvideox-2b-w4a4-unknown",
        "hub_url": "https://huggingface.co/harahan/qvgen-cogvideox-2b-w4a4",
        "weeks": [
          "2026-W36"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W36"
      ],
      "measured_on": [
        {
          "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "windows|amd64|8|16|none|0|||"
          },
          "machine_class": "windows|amd64|8|16|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "Harahan/QVGen-CogVideoX-2B-W4A4",
          "engine": null,
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-05T11:13:13.736Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/cogvideox/2b/harahan-qvgen-cogvideox-2b-w4a4-unknown",
          "hub_url": "https://huggingface.co/harahan/qvgen-cogvideox-2b-w4a4",
          "weeks": [
            "2026-W36"
          ],
          "providers": [
            "9e3c79"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Qwen-Image-2.1-Lightning",
      "made_by": "lightx2v",
      "maker_logo": null,
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Other",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/lightx2v/Qwen-Image-2.1-Lightning",
      "download_gb": null,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 15,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "linux|amd64|4|31|nvidia|15|||"
        },
        "machine_class": "linux|amd64|4|31|nvidia|15|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "lightx2v/Qwen-Image-2.1-Lightning",
        "engine": "generation",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
            "clock_ghz": 2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 1024,
            "l3_mb": 38.5,
            "single_core_hashes_per_sec": 182.6,
            "multi_core_hashes_per_sec": 387.6,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-05T03:28:40.669Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.7,
            "latency_ns": 114.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-05T03:28:42.267Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "gpu": {
            "name": "Tesla T4",
            "memory_gb": 14.6,
            "fp16_tflops": 24.5,
            "watts": 25.8,
            "tflops_per_watt": 0.947,
            "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
            "measured_at": "2026-11-05T03:49:39.140Z",
            "measured_in": "US_97058",
            "signed": true
          },
          "storage": {
            "read_mb_s": 1931.4,
            "write_mb_s": 1.2,
            "random_read_iops": 120098,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-11-05T03:29:34.741Z",
            "measured_in": "US_97058",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-21T06:30:51.338Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/qwen-image/2-1-lightning/lightx2v-qwen-image-2.1-lightning-generation",
        "hub_url": "https://huggingface.co/lightx2v/qwen-image-2.1-lightning",
        "weeks": [
          "2026-W43"
        ],
        "providers": [
          "a5be46"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "lightx2v/Qwen-Image-2.1-Lightning",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-21T06:30:51.338Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-image/2-1-lightning/lightx2v-qwen-image-2.1-lightning-generation",
          "hub_url": "https://huggingface.co/lightx2v/qwen-image-2.1-lightning",
          "weeks": [
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Sarvam-Translate (Gemma 3 4B)",
      "made_by": "sarvamai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Translation",
      "community_build": false,
      "page": "/inferences/sarvam-translate/4b",
      "hub_url": "https://huggingface.co/google/gemma-3-4b-it",
      "download_gb": 8,
      "params_b": 4,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [
        "Assamese",
        "Bengali",
        "Gujarati",
        "Hindi",
        "Kannada",
        "Malayalam",
        "Marathi",
        "Odia",
        "Punjabi",
        "Tamil",
        "Telugu",
        "Urdu",
        "Nepali"
      ],
      "status": "measured",
      "fastest": {
        "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
        "machine_detail": {
          "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "os": "Linux",
          "os_version": "ubuntu 24.04",
          "cpus": 4,
          "cpu_model": null,
          "ram_gb": 31,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "linux|amd64|4|31|none|0|||"
        },
        "machine_class": "linux|amd64|4|31|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": "BF16 (16-bit, full precision) by sarvamai",
        "ref": "sarvamai/sarvam-translate",
        "engine": "translate-hf",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Intel Xeon",
            "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
            "clock_ghz": 2.2,
            "cores_logical": 4,
            "cores_physical": 2,
            "threads_per_core": 2,
            "sockets": 1,
            "l2_kb": 256,
            "l3_mb": 55,
            "single_core_hashes_per_sec": 140.6,
            "multi_core_hashes_per_sec": 167.5,
            "sustained_ratio": 0.5,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-10-29T23:59:50.606Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "memory": {
            "total_gb": 31.3,
            "bandwidth_gb_s": 0.2,
            "latency_ns": 161.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-10-29T23:59:51.651Z",
            "measured_in": "NL_9711",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2983.7,
            "write_mb_s": 1.5,
            "random_read_iops": 281724,
            "fs_type": "overlayfs",
            "device_model": null,
            "measured_at": "2026-10-30T00:00:37.531Z",
            "measured_in": "NL_9711",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-27T14:26:11.962Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/sarvam-translate/4b/sarvamai-bf16-translate-hf",
        "hub_url": "https://huggingface.co/sarvamai/sarvam-translate",
        "weeks": [
          "2026-W44",
          "2026-W45"
        ],
        "providers": [
          "e3654f"
        ]
      },
      "cheapest": null,
      "machine_kinds": 2,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W44",
        "2026-W45"
      ],
      "measured_on": [
        {
          "machine": "Linux · no GPU · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · no GPU · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "linux|amd64|4|31|none|0|||"
          },
          "machine_class": "linux|amd64|4|31|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "BF16 (16-bit, full precision) by sarvamai",
          "ref": "sarvamai/sarvam-translate",
          "engine": "translate-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 256,
              "l3_mb": 55,
              "single_core_hashes_per_sec": 140.6,
              "multi_core_hashes_per_sec": 167.5,
              "sustained_ratio": 0.5,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-10-29T23:59:50.606Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.2,
              "latency_ns": 161.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-10-29T23:59:51.651Z",
              "measured_in": "NL_9711",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2983.7,
              "write_mb_s": 1.5,
              "random_read_iops": 281724,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-10-30T00:00:37.531Z",
              "measured_in": "NL_9711",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-27T14:26:11.962Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sarvam-translate/4b/sarvamai-bf16-translate-hf",
          "hub_url": "https://huggingface.co/sarvamai/sarvam-translate",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "e3654f"
          ]
        },
        {
          "machine": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
          "machine_detail": {
            "name": "Linux · L4 23 GB · 12 CPUs · 53 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 12,
            "cpu_model": "Intel Xeon",
            "ram_gb": 53,
            "gpu_vendor": "NVIDIA",
            "gpu_model": "L4",
            "gpu_count": 1,
            "vram_gb": 23,
            "host": "Google Colab",
            "vendor_org": "nvidia",
            "host_org": "google",
            "key": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab"
          },
          "machine_class": "linux|amd64|12|53|nvidia|23|L4|1|Google Colab",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": "BF16 (16-bit, full precision) by sarvamai",
          "ref": "sarvamai/sarvam-translate",
          "engine": "translate-hf",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.20GHz",
              "clock_ghz": 2.2,
              "cores_logical": 12,
              "cores_physical": 6,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 295.2,
              "multi_core_hashes_per_sec": 1700.6,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-07T04:46:20.920Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "memory": {
              "total_gb": 53,
              "bandwidth_gb_s": 2.8,
              "latency_ns": 23.6,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-07T04:46:21.530Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA L4",
              "memory_gb": 22,
              "fp16_tflops": 52.7,
              "watts": 34.2,
              "tflops_per_watt": 1.542,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-07T04:46:37.158Z",
              "measured_in": "SG_190464",
              "signed": true
            },
            "storage": {
              "read_mb_s": 3356.3,
              "write_mb_s": 287.7,
              "random_read_iops": 812404,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-07T04:46:22.036Z",
              "measured_in": "SG_190464",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-11-06T18:04:03.440Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/sarvam-translate/4b/sarvamai-bf16-translate-hf",
          "hub_url": "https://huggingface.co/sarvamai/sarvam-translate",
          "weeks": [
            "2026-W44",
            "2026-W45"
          ],
          "providers": [
            "dae679"
          ]
        }
      ],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "google/gemma-3-4b-it",
        "url": "https://openrouter.ai/google/gemma-3-4b-it",
        "input_usd_per_million": 0.05,
        "output_usd_per_million": 0.1,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Parasail",
            "tag": "parasail/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 0.05,
            "context_length": 131072,
            "uptime_1d": 95,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16/flex",
            "quantization": "bf16",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 0.08,
            "context_length": 131072,
            "uptime_1d": null,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.05,
            "usd_per_m_out": 0.1,
            "context_length": 131072,
            "uptime_1d": 99.6,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16/priority",
            "quantization": "bf16",
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.15,
            "context_length": 131072,
            "uptime_1d": 98.6,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 4
      },
      "licence": {
        "name": "Gemma terms of use",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/google/gemma-3-4b-it"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 2 kinds of machine here. OpenRouter also sells it."
    },
    {
      "model": "stable-audio-open-1-0-model",
      "made_by": "languaman",
      "maker_logo": null,
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Other",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/LanguaMan/stable-audio-open-1-0-model",
      "download_gb": null,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "windows|amd64|8|16|none|0|||"
        },
        "machine_class": "windows|amd64|8|16|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "LanguaMan/stable-audio-open-1-0-model",
        "engine": null,
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-08-30T01:41:15.832Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/stable-audio-open/open-1-0-model/languaman-stable-audio-open-1-0-model-unknown",
        "hub_url": "https://huggingface.co/languaman/stable-audio-open-1-0-model",
        "weeks": [
          "2026-W35"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W35"
      ],
      "measured_on": [
        {
          "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "windows|amd64|8|16|none|0|||"
          },
          "machine_class": "windows|amd64|8|16|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "LanguaMan/stable-audio-open-1-0-model",
          "engine": null,
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-08-30T01:41:15.832Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/stable-audio-open/open-1-0-model/languaman-stable-audio-open-1-0-model-unknown",
          "hub_url": "https://huggingface.co/languaman/stable-audio-open-1-0-model",
          "weeks": [
            "2026-W35"
          ],
          "providers": [
            "9e3c79"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "stable-audio-open-models",
      "made_by": "aemotionstudio",
      "maker_logo": null,
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Other",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/AEmotionStudio/stable-audio-open-models",
      "download_gb": null,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": null,
          "gpu_model": null,
          "gpu_count": 0,
          "vram_gb": null,
          "host": null,
          "vendor_org": null,
          "host_org": null,
          "key": "windows|amd64|8|16|none|0|||"
        },
        "machine_class": "windows|amd64|8|16|none|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "AEmotionStudio/stable-audio-open-models",
        "engine": null,
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-09-05T17:39:38.306Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/stable-audio-open/open-models/aemotionstudio-stable-audio-open-models-unknown",
        "hub_url": "https://huggingface.co/aemotionstudio/stable-audio-open-models",
        "weeks": [
          "2026-W36"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 1,
      "machines_total": 1,
      "weeks_measured": [
        "2026-W36"
      ],
      "measured_on": [
        {
          "machine": "Windows · no GPU · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · no GPU · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": null,
            "gpu_model": null,
            "gpu_count": 0,
            "vram_gb": null,
            "host": null,
            "vendor_org": null,
            "host_org": null,
            "key": "windows|amd64|8|16|none|0|||"
          },
          "machine_class": "windows|amd64|8|16|none|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "AEmotionStudio/stable-audio-open-models",
          "engine": null,
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-05T17:39:38.306Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/stable-audio-open/open-models/aemotionstudio-stable-audio-open-models-unknown",
          "hub_url": "https://huggingface.co/aemotionstudio/stable-audio-open-models",
          "weeks": [
            "2026-W36"
          ],
          "providers": [
            "9e3c79"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 1 kind of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "tiny-random-qwen-image",
      "made_by": "optimum-intel-internal-testing",
      "maker_logo": "https://www.gravatar.com/avatar/9d73cfd35a0ce160f557f4b4e1d6d6a1?d=retro&size=100",
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Other",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/optimum-intel-internal-testing/tiny-random-qwen-image",
      "download_gb": null,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 16,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "NVIDIA",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 12,
          "host": null,
          "vendor_org": "nvidia",
          "host_org": null,
          "key": "windows|amd64|16|16|nvidia|12|||"
        },
        "machine_class": "windows|amd64|16|16|nvidia|12|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "optimum-intel-internal-testing/tiny-random-qwen-image",
        "engine": "generation",
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "AMD Ryzen 7 5800X 8-Core",
            "model_raw": "AMD Ryzen 7 5800X 8-Core",
            "clock_ghz": null,
            "cores_logical": 16,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2085.3,
            "multi_core_hashes_per_sec": 26483.6,
            "sustained_ratio": 1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-10T09:44:03.726Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "memory": {
            "total_gb": 15.9,
            "bandwidth_gb_s": 15.3,
            "latency_ns": 11.8,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-10T09:44:04.282Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "gpu": {
            "name": "NVIDIA GeForce RTX 3080 Ti",
            "memory_gb": 12,
            "fp16_tflops": 59.8,
            "watts": null,
            "tflops_per_watt": null,
            "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
            "measured_at": "2026-11-10T09:44:16.445Z",
            "measured_in": "IN_751030",
            "signed": true
          },
          "storage": {
            "read_mb_s": 801.2,
            "write_mb_s": 73.5,
            "random_read_iops": 4296,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-10T09:44:05.469Z",
            "measured_in": "IN_751030",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-09-16T14:50:04.425Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/qwen-image/tiny-random/optimum-intel-internal-testing-tiny-random-qwen-image-generation",
        "hub_url": "https://huggingface.co/optimum-intel-internal-testing/tiny-random-qwen-image",
        "weeks": [
          "2026-W38"
        ],
        "providers": [
          "84bf54"
        ]
      },
      "cheapest": null,
      "machine_kinds": 2,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W38"
      ],
      "measured_on": [
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "optimum-intel-internal-testing/tiny-random-qwen-image",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-16T14:50:04.425Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-image/tiny-random/optimum-intel-internal-testing-tiny-random-qwen-image-generation",
          "hub_url": "https://huggingface.co/optimum-intel-internal-testing/tiny-random-qwen-image",
          "weeks": [
            "2026-W38"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "optimum-intel-internal-testing/tiny-random-qwen-image",
          "engine": "generation",
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-16T16:15:27.204Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/qwen-image/tiny-random/optimum-intel-internal-testing-tiny-random-qwen-image-generation",
          "hub_url": "https://huggingface.co/optimum-intel-internal-testing/tiny-random-qwen-image",
          "weeks": [
            "2026-W38"
          ],
          "providers": [
            "9e3c79"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 2 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Wan2.1-Fun-V1.1-1.3B-InP",
      "made_by": "alibaba-pai",
      "maker_logo": null,
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Other",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/alibaba-pai/Wan2.1-Fun-V1.1-1.3B-InP",
      "download_gb": null,
      "params_b": 1.3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "os": "Mac",
          "os_version": "macOS 26.5.2",
          "cpus": 10,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "Apple",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": null,
          "host": null,
          "vendor_org": "apple",
          "host_org": null,
          "key": "darwin|arm64|10|16|apple|0|||"
        },
        "machine_class": "darwin|arm64|10|16|apple|0|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "alibaba-pai/Wan2.1-Fun-V1.1-1.3B-InP",
        "engine": null,
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "Apple M4",
            "model_raw": "Apple M4",
            "clock_ghz": null,
            "cores_logical": 10,
            "cores_physical": 10,
            "threads_per_core": 1,
            "sockets": 1,
            "l2_kb": 4096,
            "l3_mb": null,
            "single_core_hashes_per_sec": 2850.2,
            "multi_core_hashes_per_sec": 22838.7,
            "sustained_ratio": 1.1,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-02T08:29:50.654Z",
            "measured_in": "IN_110054",
            "signed": true
          },
          "memory": {
            "total_gb": 16,
            "bandwidth_gb_s": 42.7,
            "latency_ns": 6.5,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-02T08:29:51.180Z",
            "measured_in": "IN_110054",
            "signed": true
          },
          "storage": {
            "read_mb_s": 13543.1,
            "write_mb_s": 1696,
            "random_read_iops": 1733096,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-02T08:29:51.427Z",
            "measured_in": "IN_110054",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-10-01T07:59:54.117Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/wan-2/1.3b/alibaba-pai-wan2.1-fun-v1.1-1.3b-inp-unknown",
        "hub_url": "https://huggingface.co/alibaba-pai/wan2.1-fun-v1.1-1.3b-inp",
        "weeks": [
          "2026-W40",
          "2026-W43"
        ],
        "providers": [
          "afbb8a"
        ]
      },
      "cheapest": null,
      "machine_kinds": 2,
      "machines_total": 2,
      "weeks_measured": [
        "2026-W40",
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Mac · Apple silicon · 10 CPUs · 16 GB RAM",
            "os": "Mac",
            "os_version": "macOS 26.5.2",
            "cpus": 10,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Apple",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": null,
            "host": null,
            "vendor_org": "apple",
            "host_org": null,
            "key": "darwin|arm64|10|16|apple|0|||"
          },
          "machine_class": "darwin|arm64|10|16|apple|0|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "alibaba-pai/Wan2.1-Fun-V1.1-1.3B-InP",
          "engine": null,
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Apple M4",
              "model_raw": "Apple M4",
              "clock_ghz": null,
              "cores_logical": 10,
              "cores_physical": 10,
              "threads_per_core": 1,
              "sockets": 1,
              "l2_kb": 4096,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2850.2,
              "multi_core_hashes_per_sec": 22838.7,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-02T08:29:50.654Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "memory": {
              "total_gb": 16,
              "bandwidth_gb_s": 42.7,
              "latency_ns": 6.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-02T08:29:51.180Z",
              "measured_in": "IN_110054",
              "signed": true
            },
            "storage": {
              "read_mb_s": 13543.1,
              "write_mb_s": 1696,
              "random_read_iops": 1733096,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-02T08:29:51.427Z",
              "measured_in": "IN_110054",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-01T07:59:54.117Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/wan-2/1.3b/alibaba-pai-wan2.1-fun-v1.1-1.3b-inp-unknown",
          "hub_url": "https://huggingface.co/alibaba-pai/wan2.1-fun-v1.1-1.3b-inp",
          "weeks": [
            "2026-W40",
            "2026-W43"
          ],
          "providers": [
            "afbb8a"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "alibaba-pai/Wan2.1-Fun-V1.1-1.3B-InP",
          "engine": null,
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-21T06:31:47.409Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/wan-2/1.3b/alibaba-pai-wan2.1-fun-v1.1-1.3b-inp-unknown",
          "hub_url": "https://huggingface.co/alibaba-pai/wan2.1-fun-v1.1-1.3b-inp",
          "weeks": [
            "2026-W40",
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 2 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Wan2.1-T2V-1.3B",
      "made_by": "wan-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/67b610677ea7952def8b29c6/N6jQbbeaa_FcUY-wI1dgG.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": false,
      "kind": "Other",
      "community_build": false,
      "page": null,
      "hub_url": "https://huggingface.co/Wan-AI/Wan2.1-T2V-1.3B",
      "download_gb": null,
      "params_b": 1.3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "measured",
      "fastest": {
        "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
        "machine_detail": {
          "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "os": "Windows",
          "os_version": "Windows 10.0.26200",
          "cpus": 8,
          "cpu_model": null,
          "ram_gb": 16,
          "gpu_vendor": "Intel",
          "gpu_model": null,
          "gpu_count": 1,
          "vram_gb": 2,
          "host": null,
          "vendor_org": "Intel",
          "host_org": null,
          "key": "windows|amd64|8|16|intel|2|||"
        },
        "machine_class": "windows|amd64|8|16|intel|2|||",
        "machines": 1,
        "tokens_per_sec": null,
        "context_tokens": null,
        "build": null,
        "ref": "Wan-AI/Wan2.1-T2V-1.3B",
        "engine": null,
        "cost_usd_per_million": null,
        "cost_breakdown": null,
        "machine_sheet": {
          "cpu": {
            "model": "12th Gen Intel Core i3-12100",
            "model_raw": "12th Gen Intel Core i3-12100",
            "clock_ghz": null,
            "cores_logical": 8,
            "cores_physical": null,
            "threads_per_core": null,
            "sockets": null,
            "l2_kb": null,
            "l3_mb": null,
            "single_core_hashes_per_sec": 1697.7,
            "multi_core_hashes_per_sec": 7466.7,
            "sustained_ratio": 0.9,
            "method": "sha256(1MB) single-threaded",
            "measured_at": "2026-11-15T04:07:55.704Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "memory": {
            "total_gb": 15.7,
            "bandwidth_gb_s": 4.6,
            "latency_ns": 21,
            "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
            "measured_at": "2026-11-15T04:07:56.290Z",
            "measured_in": "IN_562114",
            "signed": true
          },
          "storage": {
            "read_mb_s": 2237.9,
            "write_mb_s": 323.5,
            "random_read_iops": 149734,
            "fs_type": null,
            "device_model": null,
            "measured_at": "2026-11-15T04:07:56.778Z",
            "measured_in": "IN_562114",
            "signed": true
          }
        },
        "cost_is_estimate": false,
        "cost_basis": "the network's metered rates on this machine",
        "measured_at": "2026-09-13T16:11:32.081Z",
        "signed": true,
        "proof": null,
        "build_page": "/inferences/wan-2/1.3b/wan-ai-wan2.1-t2v-1.3b-unknown",
        "hub_url": "https://huggingface.co/wan-ai/wan2.1-t2v-1.3b",
        "weeks": [
          "2026-W36",
          "2026-W37",
          "2026-W38",
          "2026-W43"
        ],
        "providers": [
          "9e3c79"
        ]
      },
      "cheapest": null,
      "machine_kinds": 3,
      "machines_total": 3,
      "weeks_measured": [
        "2026-W36",
        "2026-W37",
        "2026-W38",
        "2026-W43"
      ],
      "measured_on": [
        {
          "machine": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · Intel 2 GB · 8 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 8,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "Intel",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 2,
            "host": null,
            "vendor_org": "Intel",
            "host_org": null,
            "key": "windows|amd64|8|16|intel|2|||"
          },
          "machine_class": "windows|amd64|8|16|intel|2|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "Wan-AI/Wan2.1-T2V-1.3B",
          "engine": null,
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "12th Gen Intel Core i3-12100",
              "model_raw": "12th Gen Intel Core i3-12100",
              "clock_ghz": null,
              "cores_logical": 8,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 1697.7,
              "multi_core_hashes_per_sec": 7466.7,
              "sustained_ratio": 0.9,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-15T04:07:55.704Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "memory": {
              "total_gb": 15.7,
              "bandwidth_gb_s": 4.6,
              "latency_ns": 21,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-15T04:07:56.290Z",
              "measured_in": "IN_562114",
              "signed": true
            },
            "storage": {
              "read_mb_s": 2237.9,
              "write_mb_s": 323.5,
              "random_read_iops": 149734,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-15T04:07:56.778Z",
              "measured_in": "IN_562114",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-13T16:11:32.081Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/wan-2/1.3b/wan-ai-wan2.1-t2v-1.3b-unknown",
          "hub_url": "https://huggingface.co/wan-ai/wan2.1-t2v-1.3b",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "9e3c79"
          ]
        },
        {
          "machine": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
          "machine_detail": {
            "name": "Windows · NVIDIA 12 GB · 16 CPUs · 16 GB RAM",
            "os": "Windows",
            "os_version": "Windows 10.0.26200",
            "cpus": 16,
            "cpu_model": null,
            "ram_gb": 16,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 12,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "windows|amd64|16|16|nvidia|12|||"
          },
          "machine_class": "windows|amd64|16|16|nvidia|12|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "Wan-AI/Wan2.1-T2V-1.3B",
          "engine": null,
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "AMD Ryzen 7 5800X 8-Core",
              "model_raw": "AMD Ryzen 7 5800X 8-Core",
              "clock_ghz": null,
              "cores_logical": 16,
              "cores_physical": null,
              "threads_per_core": null,
              "sockets": null,
              "l2_kb": null,
              "l3_mb": null,
              "single_core_hashes_per_sec": 2085.3,
              "multi_core_hashes_per_sec": 26483.6,
              "sustained_ratio": 1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-10T09:44:03.726Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "memory": {
              "total_gb": 15.9,
              "bandwidth_gb_s": 15.3,
              "latency_ns": 11.8,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-10T09:44:04.282Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "gpu": {
              "name": "NVIDIA GeForce RTX 3080 Ti",
              "memory_gb": 12,
              "fp16_tflops": 59.8,
              "watts": null,
              "tflops_per_watt": null,
              "method": "torch fp16 matmul on the card, 5 warm-up + timed iterations, 2·n³ FLOP each; pod on a time-sliced slot",
              "measured_at": "2026-11-10T09:44:16.445Z",
              "measured_in": "IN_751030",
              "signed": true
            },
            "storage": {
              "read_mb_s": 801.2,
              "write_mb_s": 73.5,
              "random_read_iops": 4296,
              "fs_type": null,
              "device_model": null,
              "measured_at": "2026-11-10T09:44:05.469Z",
              "measured_in": "IN_751030",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-09-14T05:50:54.814Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/wan-2/1.3b/wan-ai-wan2.1-t2v-1.3b-unknown",
          "hub_url": "https://huggingface.co/wan-ai/wan2.1-t2v-1.3b",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "84bf54"
          ]
        },
        {
          "machine": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
          "machine_detail": {
            "name": "Linux · NVIDIA 15 GB · 4 CPUs · 31 GB RAM",
            "os": "Linux",
            "os_version": "ubuntu 24.04",
            "cpus": 4,
            "cpu_model": null,
            "ram_gb": 31,
            "gpu_vendor": "NVIDIA",
            "gpu_model": null,
            "gpu_count": 1,
            "vram_gb": 15,
            "host": null,
            "vendor_org": "nvidia",
            "host_org": null,
            "key": "linux|amd64|4|31|nvidia|15|||"
          },
          "machine_class": "linux|amd64|4|31|nvidia|15|||",
          "machines": 1,
          "tokens_per_sec": null,
          "context_tokens": null,
          "build": null,
          "ref": "Wan-AI/Wan2.1-T2V-1.3B",
          "engine": null,
          "cost_usd_per_million": null,
          "cost_breakdown": null,
          "machine_sheet": {
            "cpu": {
              "model": "Intel Xeon",
              "model_raw": "Intel(R) Xeon(R) CPU @ 2.00GHz",
              "clock_ghz": 2,
              "cores_logical": 4,
              "cores_physical": 2,
              "threads_per_core": 2,
              "sockets": 1,
              "l2_kb": 1024,
              "l3_mb": 38.5,
              "single_core_hashes_per_sec": 182.6,
              "multi_core_hashes_per_sec": 387.6,
              "sustained_ratio": 1.1,
              "method": "sha256(1MB) single-threaded",
              "measured_at": "2026-11-05T03:28:40.669Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "memory": {
              "total_gb": 31.3,
              "bandwidth_gb_s": 0.7,
              "latency_ns": 114.5,
              "method": "sequential memcpy for bandwidth; dependent pointer chase over a buffer larger than cache for latency",
              "measured_at": "2026-11-05T03:28:42.267Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "gpu": {
              "name": "Tesla T4",
              "memory_gb": 14.6,
              "fp16_tflops": 24.5,
              "watts": 25.8,
              "tflops_per_watt": 0.947,
              "method": "torch fp16 matmul on card 0, 5 warm-up + timed iterations, 2·n³ FLOP each; host process (no cluster here)",
              "measured_at": "2026-11-05T03:49:39.140Z",
              "measured_in": "US_97058",
              "signed": true
            },
            "storage": {
              "read_mb_s": 1931.4,
              "write_mb_s": 1.2,
              "random_read_iops": 120098,
              "fs_type": "overlayfs",
              "device_model": null,
              "measured_at": "2026-11-05T03:29:34.741Z",
              "measured_in": "US_97058",
              "signed": true
            }
          },
          "cost_is_estimate": false,
          "cost_basis": "the network's metered rates on this machine",
          "measured_at": "2026-10-21T06:32:57.953Z",
          "signed": true,
          "proof": null,
          "build_page": "/inferences/wan-2/1.3b/wan-ai-wan2.1-t2v-1.3b-unknown",
          "hub_url": "https://huggingface.co/wan-ai/wan2.1-t2v-1.3b",
          "weeks": [
            "2026-W36",
            "2026-W37",
            "2026-W38",
            "2026-W43"
          ],
          "providers": [
            "a5be46"
          ]
        }
      ],
      "hosted_api": null,
      "licence": {
        "name": null,
        "you_may_serve": "Licence not checked yet",
        "class": "unknown",
        "source": null
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Runs on 3 kinds of machine here, and no hosted API sells it — this network is where to get it."
    },
    {
      "model": "Gemma 3 4B",
      "made_by": "google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/gemma-3/4b",
      "hub_url": "https://huggingface.co/google/gemma-3-4b-it",
      "download_gb": 2.3,
      "params_b": 4,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "google/gemma-3-4b-it",
        "url": "https://openrouter.ai/google/gemma-3-4b-it",
        "input_usd_per_million": 0.05,
        "output_usd_per_million": 0.1,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Parasail",
            "tag": "parasail/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 0.05,
            "context_length": 131072,
            "uptime_1d": 95,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16/flex",
            "quantization": "bf16",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 0.08,
            "context_length": 131072,
            "uptime_1d": null,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.05,
            "usd_per_m_out": 0.1,
            "context_length": 131072,
            "uptime_1d": 99.6,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16/priority",
            "quantization": "bf16",
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.15,
            "context_length": 131072,
            "uptime_1d": 98.6,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 4
      },
      "licence": {
        "name": "Gemma terms of use",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/google/gemma-3-4b-it"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.10 per million output tokens."
    },
    {
      "model": "Hunyuan MT 2 7B",
      "made_by": "tencent",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/Lp3m-XLpjQGwBItlvn69q.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/hy-mt-2/7b",
      "hub_url": "https://huggingface.co/tencent/Hy-MT2-7B",
      "download_gb": 4.3,
      "params_b": 7,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "tencent/hy-mt2-7b",
        "url": "https://openrouter.ai/tencent/hy-mt2-7b",
        "input_usd_per_million": 0.074,
        "output_usd_per_million": 0.295,
        "free_tier": false,
        "context_tokens": 8192,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Tencent",
            "tag": "tencent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.074,
            "usd_per_m_out": 0.295,
            "context_length": 8192,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/tencent/Hy-MT2-7B"
      },
      "builds": 4,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.29 per million output tokens."
    },
    {
      "model": "Ministral 3 8B-Instruct-2512",
      "made_by": "mistralai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/634c17653d11eaedd88b314d/9OgyfKstSZtbmsmuG8MbU.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/ministral-3/8b-instruct-2512",
      "hub_url": "https://huggingface.co/mistralai/Ministral-3-8B-Instruct-2512",
      "download_gb": 4.8,
      "params_b": 8,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "mistralai/ministral-8b-2512",
        "url": "https://openrouter.ai/mistralai/ministral-8b-2512",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 0.15,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.15,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 0.165,
            "usd_per_m_out": 0.165,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/mistralai/Ministral-3-8B-Instruct-2512"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.15 per million output tokens."
    },
    {
      "model": "Phi 4 Base",
      "made_by": "microsoft",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/1583646260758-5e64858c87403103f9f1055d.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/phi-4/base",
      "hub_url": "https://huggingface.co/microsoft/phi-4",
      "download_gb": 7.7,
      "params_b": 14,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "microsoft/phi-4",
        "url": "https://openrouter.ai/microsoft/phi-4",
        "input_usd_per_million": 0.07,
        "output_usd_per_million": 0.14,
        "free_tier": false,
        "context_tokens": 16384,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16/flex",
            "quantization": "bf16",
            "usd_per_m_in": 0.056,
            "usd_per_m_out": 0.112,
            "context_length": 16384,
            "uptime_1d": null,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.14,
            "context_length": 16384,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/microsoft/phi-4"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.14 per million output tokens."
    },
    {
      "model": "gpt-oss Safeguard-20B",
      "made_by": "openai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68783facef79a05727260de3/UPX5RQxiPGA-ZbBmArIKq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/gpt-oss/safeguard-20b",
      "hub_url": "https://huggingface.co/openai/gpt-oss-safeguard-20b",
      "download_gb": 10.4,
      "params_b": 20,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "openai/gpt-oss-safeguard-20b",
        "url": "https://openrouter.ai/openai/gpt-oss-safeguard-20b",
        "input_usd_per_million": 0.075,
        "output_usd_per_million": 0.3,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Groq",
            "tag": "groq",
            "quantization": null,
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.3,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openai/gpt-oss-safeguard-20b"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.30 per million output tokens."
    },
    {
      "model": "Gemma 3 27B",
      "made_by": "google",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/WtA3YYitedOr9n02eHfJe.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/gemma-3/27b",
      "hub_url": "https://huggingface.co/google/gemma-3-27b-it",
      "download_gb": 15.4,
      "params_b": 27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "google/gemma-3-27b-it",
        "url": "https://openrouter.ai/google/gemma-3-27b-it",
        "input_usd_per_million": 0.08,
        "output_usd_per_million": 0.35,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.064,
            "usd_per_m_out": 0.128,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.16,
            "context_length": 131072,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.24,
            "context_length": 131072,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.3,
            "context_length": 110000,
            "uptime_1d": 99.6,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "NextBit",
            "tag": "nextbit/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.35,
            "context_length": 131072,
            "uptime_1d": 99.4,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.45,
            "context_length": 131072,
            "uptime_1d": 99,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 6
      },
      "licence": {
        "name": "Gemma terms of use",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/google/gemma-3-27b-it"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.35 per million output tokens."
    },
    {
      "model": "Qwen 3.5 27B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3-5/27b",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.5-27B",
      "download_gb": 15.6,
      "params_b": 27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.5-27b",
        "url": "https://openrouter.ai/qwen/qwen3.5-27b",
        "input_usd_per_million": 0.26,
        "output_usd_per_million": 2.6,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.195,
            "usd_per_m_out": 1.56,
            "context_length": 262144,
            "uptime_1d": 98.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 1.75,
            "context_length": 262144,
            "uptime_1d": 97.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.25,
            "usd_per_m_out": 2,
            "context_length": 262144,
            "uptime_1d": 70.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.27,
            "usd_per_m_out": 2.16,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 2.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.26,
            "usd_per_m_out": 2.6,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 6
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.5-27B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $2.60 per million output tokens."
    },
    {
      "model": "Qwen 3 30B-A3B-Instruct-2507",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3/30b-a3b-instruct-2507",
      "hub_url": "https://huggingface.co/Qwen/Qwen3-30B-A3B-Instruct-2507",
      "download_gb": 17.3,
      "params_b": 30,
      "active_params_b": 3,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3-30b-a3b-instruct-2507",
        "url": "https://openrouter.ai/qwen/qwen3-30b-a3b-instruct-2507",
        "input_usd_per_million": 0.1,
        "output_usd_per_million": 0.3,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.09,
            "usd_per_m_out": 0.3,
            "context_length": 262144,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.09,
            "usd_per_m_out": 0.3,
            "context_length": 262144,
            "uptime_1d": 97.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.3,
            "context_length": 262144,
            "uptime_1d": 94.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 3
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3-30B-A3B-Instruct-2507"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.30 per million output tokens."
    },
    {
      "model": "Qwen 3 Coder-Next",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3/coder-next",
      "hub_url": "https://huggingface.co/Qwen/Qwen3-Coder-Next",
      "download_gb": 41.8,
      "params_b": 74,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3-coder-next",
        "url": "https://openrouter.ai/qwen/qwen3-coder-next",
        "input_usd_per_million": 0.12,
        "output_usd_per_million": 0.8,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Parasail",
            "tag": "parasail/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.8,
            "context_length": 262144,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3-Coder-Next-FP8"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.80 per million output tokens."
    },
    {
      "model": "GLM-4.6 V",
      "made_by": "zai-org",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/glm-4-6/v",
      "hub_url": "https://huggingface.co/zai-org/GLM-4.6V",
      "download_gb": 54.5,
      "params_b": 97,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-4.6v",
        "url": "https://openrouter.ai/z-ai/glm-4.6v",
        "input_usd_per_million": 0.3,
        "output_usd_per_million": 0.9,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Novita",
            "tag": "novita/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 0.9,
            "context_length": 131072,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 0.9,
            "context_length": 131072,
            "uptime_1d": 91.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-4.6V"
      },
      "builds": 6,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.90 per million output tokens."
    },
    {
      "model": "GLM-4.5 AIR",
      "made_by": "zai-org",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/glm-4-5/air",
      "hub_url": "https://huggingface.co/zai-org/GLM-4.5-Air",
      "download_gb": 56,
      "params_b": 99,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-4.5-air",
        "url": "https://openrouter.ai/z-ai/glm-4.5-air",
        "input_usd_per_million": 0.2,
        "output_usd_per_million": 1.1,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Novita",
            "tag": "novita/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.85,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.86,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 1.1,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 3
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-4.5-Air"
      },
      "builds": 3,
      "variants": [
        {
          "model": "GLM-4.5-Air-mixed_4_6",
          "made_by": "terion-mlx",
          "page": "/inferences/glm-4-5/air~terion-mlx-mixed-4-6",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.10 per million output tokens."
    },
    {
      "model": "gpt-oss 120B",
      "made_by": "openai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/68783facef79a05727260de3/UPX5RQxiPGA-ZbBmArIKq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/gpt-oss/120b",
      "hub_url": "https://huggingface.co/openai/gpt-oss-120b",
      "download_gb": 58.5,
      "params_b": 120,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "openai/gpt-oss-120b",
        "url": "https://openrouter.ai/openai/gpt-oss-120b",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 0.6,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.035,
            "usd_per_m_out": 0.075,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.15,
            "context_length": 131072,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.15,
            "context_length": 128000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.17,
            "context_length": 131072,
            "uptime_1d": 91.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.037,
            "usd_per_m_out": 0.17,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.18,
            "context_length": 131072,
            "uptime_1d": 99.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.187,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.05,
            "usd_per_m_out": 0.25,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.25,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Google",
            "tag": "google-vertex/global",
            "quantization": null,
            "usd_per_m_in": 0.09,
            "usd_per_m_out": 0.36,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.4,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.42,
            "context_length": 128000,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Io Net",
            "tag": "io-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.25,
            "usd_per_m_out": 0.45,
            "context_length": 131052,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp4/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.5,
            "context_length": 128072,
            "uptime_1d": 98.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.5,
            "context_length": 128072,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock/eu-west-1",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 131072,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 131072,
            "uptime_1d": 90.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Groq",
            "tag": "groq",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 131072,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.75,
            "context_length": 131072,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.75,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cerebras",
            "tag": "cerebras/fp16",
            "quantization": "fp16",
            "usd_per_m_in": 0.35,
            "usd_per_m_out": 0.75,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SambaNova",
            "tag": "sambanova",
            "quantization": null,
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.95,
            "context_length": 131072,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 0.95,
            "context_length": 131072,
            "uptime_1d": 60.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 26
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/openai/gpt-oss-120b"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.60 per million output tokens."
    },
    {
      "model": "Devstral 2 123B-Instruct-2512",
      "made_by": "mistralai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/634c17653d11eaedd88b314d/9OgyfKstSZtbmsmuG8MbU.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/devstral-2/123b-instruct-2512",
      "hub_url": "https://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512",
      "download_gb": 65.5,
      "params_b": 123,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "mistralai/devstral-2512",
        "url": "https://openrouter.ai/mistralai/devstral-2512",
        "input_usd_per_million": 0.4,
        "output_usd_per_million": 2,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 0.4,
            "usd_per_m_out": 2,
            "context_length": 262144,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/mistralai/Devstral-2-123B-Instruct-2512"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $2.00 per million output tokens."
    },
    {
      "model": "Mistral Small 4 119B-2603",
      "made_by": "mistralai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/634c17653d11eaedd88b314d/9OgyfKstSZtbmsmuG8MbU.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/mistral-small-4/119b-2603",
      "hub_url": "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603",
      "download_gb": 67.2,
      "params_b": 119,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "mistralai/mistral-small-2603",
        "url": "https://openrouter.ai/mistralai/mistral-small-2603",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 0.6,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 0.165,
            "usd_per_m_out": 0.66,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/us",
            "quantization": null,
            "usd_per_m_in": 0.165,
            "usd_per_m_out": 0.66,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 3
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Ares-Realm-Studios/Mistral-Small-4-119B-2603"
      },
      "builds": 8,
      "variants": [
        {
          "model": "Mistral-Small-4-119B-2603-Mixed-3bit-MLX",
          "made_by": "exalandru",
          "page": "/inferences/mistral-small-4/119b-2603~exalandru-mixed",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "Mistral-Small-4-119B-TurboQuant-MLX-4bit",
          "made_by": "majentik",
          "page": "/inferences/mistral-small-4/119b-2603~majentik-turboquant",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.60 per million output tokens."
    },
    {
      "model": "Mistral Medium 3.5 128B",
      "made_by": "mistralai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/634c17653d11eaedd88b314d/9OgyfKstSZtbmsmuG8MbU.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/mistral-medium-3-5/128b",
      "hub_url": "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B",
      "download_gb": 69.1,
      "params_b": 128,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "mistralai/mistral-medium-3-5",
        "url": "https://openrouter.ai/mistralai/mistral-medium-3-5",
        "input_usd_per_million": 1.5,
        "output_usd_per_million": 7.5,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": null,
            "usd_per_m_in": 1.5,
            "usd_per_m_out": 7.5,
            "context_length": 262144,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 1.65,
            "usd_per_m_out": 8.25,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B"
      },
      "builds": 7,
      "variants": [
        {
          "model": "4xH100-Mistral-Medium-3.5-128B-MonEspaceSante",
          "made_by": "fenyo",
          "page": "/inferences/mistral-medium-3-5/128b~fenyo-4xh100-monespacesante",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "Mistral-Medium-3.5-128B-Eschaton-Uncensored-GGUF",
          "made_by": "cloudbjorn",
          "page": "/inferences/mistral-medium-3-5/128b~cloudbjorn-eschaton-uncensored",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $7.50 per million output tokens."
    },
    {
      "model": "Step 3.7 Flash",
      "made_by": "stepfun-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/644f7e6233ac8f46fa0b9e26/CmF2ocXhkr2UtHXgmwq7-.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/step-3-7/flash",
      "hub_url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash",
      "download_gb": 103.2,
      "params_b": 183,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "stepfun/step-3.7-flash",
        "url": "https://openrouter.ai/stepfun/step-3.7-flash",
        "input_usd_per_million": 0.2,
        "output_usd_per_million": 1.15,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 1.15,
            "context_length": 262144,
            "uptime_1d": 99.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StepFun",
            "tag": "stepfun/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 1.15,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/stepfun-ai/Step-3.7-Flash"
      },
      "builds": 6,
      "variants": [
        {
          "model": "Step-3.7-173B-Ream-4.6bit-MLX",
          "made_by": "True2456",
          "page": "/inferences/step-3-7/flash~true2456-ream-4",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "Step-3.7-173B-Reap-4.6bit-MLX",
          "made_by": "True2456",
          "page": "/inferences/step-3-7/flash~true2456-reap-4",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.15 per million output tokens."
    },
    {
      "model": "MiniMax-M2.5 Base",
      "made_by": "MiniMaxAI",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/676e38ad04af5bec20bc9faf/dUd-LsZEX0H_d4qefO_g6.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/minimax-m2-5/base",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M2.5",
      "download_gb": 119.8,
      "params_b": 212,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m2.5",
        "url": "https://openrouter.ai/minimax/minimax-m2.5",
        "input_usd_per_million": 0.27,
        "output_usd_per_million": 0.95,
        "free_tier": false,
        "context_tokens": 198000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 0.27,
            "usd_per_m_out": 0.95,
            "context_length": 198000,
            "uptime_1d": 98.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.295,
            "usd_per_m_out": 1.2,
            "context_length": 196608,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 196608,
            "uptime_1d": 98.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 65536,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 4
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/MiniMaxAI/MiniMax-M2.5"
      },
      "builds": 5,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.95 per million output tokens."
    },
    {
      "model": "MiniMax-M2.7 Base",
      "made_by": "MiniMaxAI",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/676e38ad04af5bec20bc9faf/dUd-LsZEX0H_d4qefO_g6.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/minimax-m2-7/base",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7",
      "download_gb": 119.8,
      "params_b": 212,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m2.7",
        "url": "https://openrouter.ai/minimax/minimax-m2.7",
        "input_usd_per_million": 0.21,
        "output_usd_per_million": 0.84,
        "free_tier": false,
        "context_tokens": 196608,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 196608,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Groq",
            "tag": "groq",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.8,
            "context_length": 196608,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7"
      },
      "builds": 6,
      "variants": [
        {
          "model": "MiniMax-M2.7-Abliterated-Heretic-GGUF",
          "made_by": "Youssofal",
          "page": "/inferences/minimax-m2-7/base~youssofal-abliterated-heretic",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "MiniMax-M2.7-RAM-100GB-MLX",
          "made_by": "baa-ai",
          "page": "/inferences/minimax-m2-7/base~baa-ai-ram-100gb",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "MiniMax-M2.7-RAM-120GB-MLX",
          "made_by": "baa-ai",
          "page": "/inferences/minimax-m2-7/base~baa-ai-ram-120gb",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "MiniMax-M2.7-RAM-91GB-MLX",
          "made_by": "baa-ai",
          "page": "/inferences/minimax-m2-7/base~baa-ai-ram-91gb",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "MiniMax-M2.7-TQ3",
          "made_by": "fno2010",
          "page": "/inferences/minimax-m2-7/base~fno2010-tq3",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "MiniMax-M2.7-ConfigI-MLX",
          "made_by": "thetom-ai",
          "page": "/inferences/minimax-m2-7/base~thetom-ai-configi",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.84 per million output tokens."
    },
    {
      "model": "Hunyuan Hy3 Preview",
      "made_by": "tencent",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/Lp3m-XLpjQGwBItlvn69q.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/hy-3/preview",
      "hub_url": "https://huggingface.co/tencent/Hy3-preview",
      "download_gb": 154.6,
      "params_b": 274,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "tencent/hy3-preview",
        "url": "https://openrouter.ai/tencent/hy3-preview",
        "input_usd_per_million": 0.18,
        "output_usd_per_million": 0.6,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "GMICloud",
            "tag": "gmicloud/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.6,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/tencent/Hy3-preview"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.60 per million output tokens."
    },
    {
      "model": "DeepSeek V4 Flash-0731",
      "made_by": "deepseek-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/deepseek-v4/flash-0731",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731",
      "download_gb": 155.4,
      "params_b": 157,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v4-flash-0731",
        "url": "https://openrouter.ai/deepseek/deepseek-v4-flash-0731",
        "input_usd_per_million": 0.0434,
        "output_usd_per_million": 0.3,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.14,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.048,
            "usd_per_m_out": 0.144,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.18,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/fast",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.23,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.119,
            "usd_per_m_out": 0.238,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.26,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp32",
            "quantization": "fp32",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.26,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.26,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.088,
            "usd_per_m_out": 0.264,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.09,
            "usd_per_m_out": 0.27,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.27,
            "context_length": 262144,
            "uptime_1d": 92.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 98.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.0434,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.42,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.495,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.176,
            "usd_per_m_out": 0.528,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/us-east-1",
            "quantization": null,
            "usd_per_m_in": 0.187,
            "usd_per_m_out": 0.561,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.22,
            "usd_per_m_out": 0.66,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.286,
            "usd_per_m_out": 0.858,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.4092,
            "usd_per_m_out": 1.2276,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.44,
            "usd_per_m_out": 1.32,
            "context_length": 1048576,
            "uptime_1d": null,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.44,
            "usd_per_m_out": 1.32,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.44,
            "usd_per_m_out": 1.32,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.44,
            "usd_per_m_out": 1.32,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.44,
            "usd_per_m_out": 1.32,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 29
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731"
      },
      "builds": 3,
      "variants": [
        {
          "model": "DeepSeek-V4-Flash-0731-abliterated",
          "made_by": "prem-research",
          "page": "/inferences/deepseek-v4/flash-0731~prem-research-abliterated",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.30 per million output tokens."
    },
    {
      "model": "MiMo V2.6 Flash-RL",
      "made_by": "XiaomiMiMo",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/680cb7d1233834890a64acee/5w_4aLfF-7MAyaIPOV498.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/mimo-v2-6/flash-rl",
      "hub_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "download_gb": 155.8,
      "params_b": 276,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "xiaomi/mimo-v2.6-flash",
        "url": "https://openrouter.ai/xiaomi/mimo-v2.6-flash",
        "input_usd_per_million": 0.14,
        "output_usd_per_million": 0.28,
        "free_tier": false,
        "context_tokens": 1050000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "GMICloud",
            "tag": "gmicloud/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.112,
            "usd_per_m_out": 0.224,
            "context_length": 1050000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 98.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Xiaomi",
            "tag": "xiaomi/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Tenstorrent",
            "tag": "tenstorrent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.165,
            "usd_per_m_out": 0.33,
            "context_length": 1048576,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.35,
            "context_length": 1000000,
            "uptime_1d": 97,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.36,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 8
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL"
      },
      "builds": 3,
      "variants": [
        {
          "model": "MiMo-V2.6-Flash-RL-Uncensored",
          "made_by": "dealignai",
          "page": "/inferences/mimo-v2-6/flash-rl~dealignai-uncensored",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.28 per million output tokens."
    },
    {
      "model": "DeepSeek V4 Flash-Vision-EXP",
      "made_by": "deepseek-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/deepseek-v4/flash-vision-exp",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-Vision-Exp",
      "download_gb": 156.3,
      "params_b": 158,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v4-flash-vision-exp",
        "url": "https://openrouter.ai/deepseek/deepseek-v4-flash-vision-exp",
        "input_usd_per_million": 0.44,
        "output_usd_per_million": 1.32,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.17248,
            "usd_per_m_out": 0.51744,
            "context_length": 1048576,
            "uptime_1d": 66.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2156,
            "usd_per_m_out": 0.6468,
            "context_length": 1048576,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.3234,
            "usd_per_m_out": 0.9702,
            "context_length": 1048576,
            "uptime_1d": 98.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.44,
            "usd_per_m_out": 1.32,
            "context_length": 1048575,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 4
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-Vision-Exp"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.32 per million output tokens."
    },
    {
      "model": "DeepSeek V4 Flash",
      "made_by": "deepseek-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/deepseek-v4/flash",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash",
      "download_gb": 156.7,
      "params_b": 158,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v4-flash",
        "url": "https://openrouter.ai/deepseek/deepseek-v4-flash",
        "input_usd_per_million": 0.09,
        "output_usd_per_million": 0.18,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.072,
            "usd_per_m_out": 0.144,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.084,
            "usd_per_m_out": 0.168,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.17,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.09,
            "usd_per_m_out": 0.18,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.091,
            "usd_per_m_out": 0.182,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 0.0966,
            "usd_per_m_out": 0.1925,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.105,
            "usd_per_m_out": 0.21,
            "context_length": 1048576,
            "uptime_1d": 93.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.135,
            "usd_per_m_out": 0.27,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 92,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.014,
            "usd_per_m_out": 0.45,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Azure",
            "tag": "azure/us",
            "quantization": null,
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.56,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 15
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.18 per million output tokens."
    },
    {
      "model": "Hunyuan Hy3 Base",
      "made_by": "tencent",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/Lp3m-XLpjQGwBItlvn69q.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/hy-3/base",
      "hub_url": "https://huggingface.co/tencent/Hy3",
      "download_gb": 169.6,
      "params_b": 300,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "tencent/hy3",
        "url": "https://openrouter.ai/tencent/hy3",
        "input_usd_per_million": 0.0825,
        "output_usd_per_million": 0.33,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Tencent",
            "tag": "tencent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.0825,
            "usd_per_m_out": 0.33,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.58,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.58,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 0.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 4
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/tencent/Hy3"
      },
      "builds": 4,
      "variants": [
        {
          "model": "Hy3-Alis-MLX-Dynamic",
          "made_by": "avlp12",
          "page": "/inferences/hy-3/base~avlp12-alis-dynamic",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "hy3-tiny-random",
          "made_by": "yujiepan",
          "page": "/inferences/hy-3/base~yujiepan-tiny-random",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.33 per million output tokens."
    },
    {
      "model": "Qwen 3.8 Flash-Next",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/qwen-3-8/flash-next",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next",
      "download_gb": 172.8,
      "params_b": 175,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.8-flash",
        "url": "https://openrouter.ai/qwen/qwen3.8-flash",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 0.47,
        "free_tier": false,
        "context_tokens": 1000000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Alibaba",
            "tag": "alibaba/us-east-1",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.47,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.47,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Qwen Community 1.0 licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next"
      },
      "builds": 3,
      "variants": [
        {
          "model": "Qwen3.8-Flash-Next-Reap-288-MLX-4bit",
          "made_by": "sh0wie",
          "page": "/inferences/qwen-3-8/flash-next~sh0wie-reap-288",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.47 per million output tokens."
    },
    {
      "model": "MiMo V2.5 Base",
      "made_by": "XiaomiMiMo",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/680cb7d1233834890a64acee/5w_4aLfF-7MAyaIPOV498.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/mimo-v2-5/base",
      "hub_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.5",
      "download_gb": 177.8,
      "params_b": 315,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "xiaomi/mimo-v2.5",
        "url": "https://openrouter.ai/xiaomi/mimo-v2.5",
        "input_usd_per_million": 0.4,
        "output_usd_per_million": 2,
        "free_tier": false,
        "context_tokens": 1000000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.4,
            "usd_per_m_out": 2,
            "context_length": 1000000,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.5"
      },
      "builds": 3,
      "variants": [
        {
          "model": "MiMo-V2.5-Text-Only-oQ4-MLX",
          "made_by": "Blightbow",
          "page": "/inferences/mimo-v2-5/base~blightbow-text-only-oq4",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $2.00 per million output tokens."
    },
    {
      "model": "GLM-4.5 Base",
      "made_by": "zai-org",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/glm-4-5/base",
      "hub_url": "https://huggingface.co/zai-org/GLM-4.5",
      "download_gb": 184.9,
      "params_b": 328,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-4.5",
        "url": "https://openrouter.ai/z-ai/glm-4.5",
        "input_usd_per_million": 0.6,
        "output_usd_per_million": 2.2,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.2,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-4.5"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $2.20 per million output tokens."
    },
    {
      "model": "GLM-4.6 Base",
      "made_by": "zai-org",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/glm-4-6/base",
      "hub_url": "https://huggingface.co/zai-org/GLM-4.6",
      "download_gb": 184.9,
      "params_b": 328,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-4.6",
        "url": "https://openrouter.ai/z-ai/glm-4.6",
        "input_usd_per_million": 0.5,
        "output_usd_per_million": 2,
        "free_tier": false,
        "context_tokens": 204800,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/flex",
            "quantization": "fp4",
            "usd_per_m_in": 0.4,
            "usd_per_m_out": 1.6,
            "context_length": 202752,
            "uptime_1d": null,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.43,
            "usd_per_m_out": 1.75,
            "context_length": 198000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.5,
            "usd_per_m_out": 2,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.55,
            "usd_per_m_out": 2.2,
            "context_length": 204800,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.2,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 5
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-4.6"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $2.00 per million output tokens."
    },
    {
      "model": "GLM-4.7 Base",
      "made_by": "zai-org",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/glm-4-7/base",
      "hub_url": "https://huggingface.co/zai-org/GLM-4.7",
      "download_gb": 184.9,
      "params_b": 328,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-4.7",
        "url": "https://openrouter.ai/z-ai/glm-4.7",
        "input_usd_per_million": 0.4,
        "output_usd_per_million": 1.75,
        "free_tier": false,
        "context_tokens": 204800,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/flex",
            "quantization": "fp4",
            "usd_per_m_in": 0.32,
            "usd_per_m_out": 1.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.4,
            "usd_per_m_out": 1.75,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.4004,
            "usd_per_m_out": 1.9292,
            "context_length": 198000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.2,
            "context_length": 204800,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.2,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 2.5,
            "context_length": 131072,
            "uptime_1d": 99.8,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 6
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-4.7"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.75 per million output tokens."
    },
    {
      "model": "MiniMax-M3 Base",
      "made_by": "MiniMaxAI",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/676e38ad04af5bec20bc9faf/dUd-LsZEX0H_d4qefO_g6.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/minimax-m3/base",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M3",
      "download_gb": 224.9,
      "params_b": 398,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m3",
        "url": "https://openrouter.ai/minimax/minimax-m3",
        "input_usd_per_million": 0.3,
        "output_usd_per_million": 1.2,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.224,
            "usd_per_m_out": 0.88,
            "context_length": 524288,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.23,
            "usd_per_m_out": 0.96,
            "context_length": 262144,
            "uptime_1d": 91.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 0.96,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 1.1,
            "context_length": 524288,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 524288,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Minimax",
            "tag": "minimax",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 524288,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 8
      },
      "licence": {
        "name": "Minimax Community licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/MiniMaxAI/MiniMax-M3"
      },
      "builds": 4,
      "variants": [
        {
          "model": "MiniMax-M3-Mixed-4.5bit-MLX",
          "made_by": "Spangler3000",
          "page": "/inferences/minimax-m3/base~spangler3000-mixed-4",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "MiniMax-M3-DSpark-GGUF",
          "made_by": "Anbeeld",
          "page": "/inferences/minimax-m3/base~anbeeld-dspark",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "MiniMax-M3-ConfigI-MLX",
          "made_by": "thetom-ai",
          "page": "/inferences/minimax-m3/base~thetom-ai-configi",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.20 per million output tokens."
    },
    {
      "model": "MiniMax-M3.1 Flash",
      "made_by": "MiniMaxAI",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/676e38ad04af5bec20bc9faf/dUd-LsZEX0H_d4qefO_g6.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/minimax-m3-1/flash",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash",
      "download_gb": 231.7,
      "params_b": 410,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m3.1-flash",
        "url": "https://openrouter.ai/minimax/minimax-m3.1-flash",
        "input_usd_per_million": 0.075,
        "output_usd_per_million": 0.3,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Inferact vLLM",
            "tag": "inferact-vllm/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Minimax",
            "tag": "minimax",
            "quantization": null,
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.135,
            "usd_per_m_out": 0.54,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.135,
            "usd_per_m_out": 0.54,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SambaNova",
            "tag": "sambanova",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp4/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1000000,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 13
      },
      "licence": {
        "name": "Minimax Community licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/LyJonathan/MiniMax-M3.1-Flash-GGUF"
      },
      "builds": 2,
      "variants": [
        {
          "model": "Sakura-MiniMax-M3.1-Flash-Q3_K_M-GGUF",
          "made_by": "webmp3",
          "page": "/inferences/minimax-m3-1/flash~webmp3-sakura-k-m",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.30 per million output tokens."
    },
    {
      "model": "GLM-5.3 Flash",
      "made_by": "zai-org",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/glm-5-3/flash",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.3-Flash",
      "download_gb": 305.8,
      "params_b": 309,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.3-flash",
        "url": "https://openrouter.ai/z-ai/glm-5.3-flash",
        "input_usd_per_million": 0.0399,
        "output_usd_per_million": 0.0784,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.0399,
            "usd_per_m_out": 0.0784,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.065,
            "usd_per_m_out": 0.18,
            "context_length": 262144,
            "uptime_1d": 96.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.174,
            "usd_per_m_out": 0.183,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.042,
            "usd_per_m_out": 0.184,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.065,
            "usd_per_m_out": 0.24,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.24,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Tenstorrent",
            "tag": "tenstorrent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.25,
            "context_length": 262000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.084,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.0885,
            "usd_per_m_out": 0.295,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "OpenInference",
            "tag": "open-inference/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.108,
            "usd_per_m_out": 0.35,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.108,
            "usd_per_m_out": 0.35,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.1125,
            "usd_per_m_out": 0.375,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.38,
            "context_length": 1048576,
            "uptime_1d": 90.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.4,
            "context_length": 1048576,
            "uptime_1d": 98.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "NextBit",
            "tag": "nextbit/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.45,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.47,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Particle AI",
            "tag": "particle-ai/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "KT",
            "tag": "kt",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 96.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.625,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.625,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp4",
            "usd_per_m_in": 0.1875,
            "usd_per_m_out": 0.625,
            "context_length": 1048576,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 0.75,
            "context_length": 1048576,
            "uptime_1d": 95,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 42
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.3-Flash"
      },
      "builds": 5,
      "variants": [
        {
          "model": "GLM-5.3-Flash-REAP50-GGUF",
          "made_by": "patrickbdevaney",
          "page": "/inferences/glm-5-3/flash~patrickbdevaney-reap50",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "GLM-5.3-Flash-Coder-RCO-152E-GGUF",
          "made_by": "lapadatbogdan",
          "page": "/inferences/glm-5-3/flash~lapadatbogdan-coder-rco-152e",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "GLM-5.3-Flash-MLX-mixed-4_8bit",
          "made_by": "pipenetwork",
          "page": "/inferences/glm-5-3/flash~pipenetwork-mixed-4",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not run here yet: it needs about 305.8 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $0.08 per million output tokens."
    },
    {
      "model": "DeepSeek V3.1 Base",
      "made_by": "deepseek-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/deepseek-v3-1/base",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V3.1",
      "download_gb": 351.7,
      "params_b": 623,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-chat-v3.1",
        "url": "https://openrouter.ai/deepseek/deepseek-chat-v3.1",
        "input_usd_per_million": 0.25,
        "output_usd_per_million": 0.95,
        "free_tier": false,
        "context_tokens": 163840,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.25,
            "usd_per_m_out": 0.95,
            "context_length": 163840,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.27,
            "usd_per_m_out": 1,
            "context_length": 163840,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.5,
            "usd_per_m_out": 1.5,
            "context_length": 161000,
            "uptime_1d": 91.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 3
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V3.1"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 351.7 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $0.95 per million output tokens."
    },
    {
      "model": "DeepSeek V3.1 Terminus",
      "made_by": "deepseek-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/deepseek-v3-1/terminus",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V3.1-Terminus",
      "download_gb": 351.7,
      "params_b": 623,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v3.1-terminus",
        "url": "https://openrouter.ai/deepseek/deepseek-v3.1-terminus",
        "input_usd_per_million": 0.27,
        "output_usd_per_million": 1,
        "free_tier": false,
        "context_tokens": 163840,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.27,
            "usd_per_m_out": 1,
            "context_length": 163840,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V3.1-Terminus"
      },
      "builds": 2,
      "variants": [
        {
          "model": "DeepSeek-V3.1-Terminus-MLX-mixed_4_6",
          "made_by": "finding1",
          "page": "/inferences/deepseek-v3-1/terminus~finding1-mixed-4-6",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "DeepSeek-V3.1-Terminus-MLX-mixed_4_6-v0.28.2",
          "made_by": "finding1",
          "page": "/inferences/deepseek-v3-1/terminus~finding1-mixed-4-6-28-2",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not run here yet: it needs about 351.7 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $1.00 per million output tokens."
    },
    {
      "model": "GLM-5.2 Base",
      "made_by": "zai-org",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/glm-5-2/base",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.2-FP8",
      "download_gb": 368,
      "params_b": 652,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.2",
        "url": "https://openrouter.ai/z-ai/glm-5.2",
        "input_usd_per_million": 0.007,
        "output_usd_per_million": 6.8,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 0.88,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/flex",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.44,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.5625,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.643,
            "usd_per_m_out": 1.978,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ambient",
            "tag": "ambient/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2,
            "context_length": 202752,
            "uptime_1d": 50.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6496,
            "usd_per_m_out": 2.0416,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.651,
            "usd_per_m_out": 2.046,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 2.2,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.62,
            "usd_per_m_out": 2.42,
            "context_length": 1048576,
            "uptime_1d": 92.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/priority",
            "quantization": "fp4",
            "usd_per_m_in": 0.84375,
            "usd_per_m_out": 2.7,
            "context_length": 1048576,
            "uptime_1d": 98.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.938,
            "usd_per_m_out": 2.948,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.966,
            "usd_per_m_out": 3.036,
            "context_length": 1000000,
            "uptime_1d": 97.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8/us-east-1",
            "quantization": "fp8",
            "usd_per_m_in": 1.12,
            "usd_per_m_out": 3.52,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": "nvfp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8/us",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 262144,
            "uptime_1d": 98.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 1.54,
            "usd_per_m_out": 4.84,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/us/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.007,
            "usd_per_m_out": 6.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.31,
            "usd_per_m_out": 7.26,
            "context_length": 1000000,
            "uptime_1d": 92.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fast",
            "quantization": "fp4",
            "usd_per_m_in": 2.25,
            "usd_per_m_out": 7.88,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.0066,
            "usd_per_m_out": 7.9833,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer",
            "quantization": "fp4",
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 34
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.2-FP8"
      },
      "builds": 3,
      "variants": [
        {
          "model": "GLM-5.2-Alis-MLX-Dynamic-3.5bpw",
          "made_by": "avlp12",
          "page": "/inferences/glm-5-2/base~avlp12-alis-dynamic-3-5bpw",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "GLM-5.2-DQ4plus-q8",
          "made_by": "mlx-community",
          "page": "/inferences/glm-5-2/base~mlx-community-dq4plus",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "GLM-5.2-Demolition-q4a4-soul-MLX",
          "made_by": "philipjohnbasile",
          "page": "/inferences/glm-5-2/base~philipjohnbasile-demolition-soul",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "GLM-5.2-REAP50-MLX-4bit",
          "made_by": "pipenetwork",
          "page": "/inferences/glm-5-2/base~pipenetwork-reap50",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not run here yet: it needs about 368 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $6.80 per million output tokens."
    },
    {
      "model": "DeepSeek V3.2 Base",
      "made_by": "deepseek-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/deepseek-v3-2/base",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V3.2",
      "download_gb": 377.6,
      "params_b": 669,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v3.2",
        "url": "https://openrouter.ai/deepseek/deepseek-v3.2",
        "input_usd_per_million": 0.26,
        "output_usd_per_million": 0.38,
        "free_tier": false,
        "context_tokens": 163840,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2088,
            "usd_per_m_out": 0.3096,
            "context_length": 163840,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.26,
            "usd_per_m_out": 0.38,
            "context_length": 163840,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.26,
            "usd_per_m_out": 0.38,
            "context_length": 163840,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 0.42,
            "context_length": 131072,
            "uptime_1d": 99.3,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.259,
            "usd_per_m_out": 0.8,
            "context_length": 163840,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 0.96,
            "context_length": 163840,
            "uptime_1d": 98.5,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 6
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V3.2"
      },
      "builds": 4,
      "variants": [
        {
          "model": "DeepSeek-V3.2-RAM-350GB-MLX",
          "made_by": "baa-ai",
          "page": "/inferences/deepseek-v3-2/base~baa-ai-ram-350gb",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not run here yet: it needs about 377.6 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $0.38 per million output tokens."
    },
    {
      "model": "Mistral Large 3 675B-Instruct-2512",
      "made_by": "mistralai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/634c17653d11eaedd88b314d/9OgyfKstSZtbmsmuG8MbU.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/mistral-large-3/675b-instruct-2512",
      "hub_url": "https://huggingface.co/mistralai/Mistral-Large-3-675B-Instruct-2512",
      "download_gb": 377.9,
      "params_b": 675,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "mistralai/mistral-large-2512",
        "url": "https://openrouter.ai/mistralai/mistral-large-2512",
        "input_usd_per_million": 0.5,
        "output_usd_per_million": 1.5,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": null,
            "usd_per_m_in": 0.5,
            "usd_per_m_out": 1.5,
            "context_length": 262144,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 0.55,
            "usd_per_m_out": 1.65,
            "context_length": 262144,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/mistralai/Mistral-Large-3-675B-Instruct-2512"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 377.9 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $1.50 per million output tokens."
    },
    {
      "model": "GLM-5 Base",
      "made_by": "zai-org",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/glm-5/base",
      "hub_url": "https://huggingface.co/zai-org/GLM-5",
      "download_gb": 389.9,
      "params_b": 691,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5",
        "url": "https://openrouter.ai/z-ai/glm-5",
        "input_usd_per_million": 1,
        "output_usd_per_million": 3.2,
        "free_tier": false,
        "context_tokens": 204800,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.92,
            "context_length": 198000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.92,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 2.24,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 2.55,
            "context_length": 204800,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock",
            "quantization": null,
            "usd_per_m_in": 1,
            "usd_per_m_out": 3.2,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1,
            "usd_per_m_out": 3.2,
            "context_length": 198000,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1,
            "usd_per_m_out": 3.2,
            "context_length": 202800,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1,
            "usd_per_m_out": 3.2,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/turbo",
            "quantization": null,
            "usd_per_m_in": 1.2,
            "usd_per_m_out": 4,
            "context_length": 200000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 9
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 389.9 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $3.20 per million output tokens."
    },
    {
      "model": "GLM-5.1 Base",
      "made_by": "zai-org",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/glm-5-1/base",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.1",
      "download_gb": 389.9,
      "params_b": 691,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.1",
        "url": "https://openrouter.ai/z-ai/glm-5.1",
        "input_usd_per_million": 1.4,
        "output_usd_per_million": 4.4,
        "free_tier": false,
        "context_tokens": 204800,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.9,
            "usd_per_m_out": 2.9,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.966,
            "usd_per_m_out": 3.036,
            "context_length": 200000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.19,
            "usd_per_m_out": 3.74,
            "context_length": 204800,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.26,
            "usd_per_m_out": 3.96,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.33,
            "usd_per_m_out": 4.18,
            "context_length": 202745,
            "uptime_1d": 95.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.38,
            "usd_per_m_out": 4.4,
            "context_length": 204800,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4014,
            "usd_per_m_out": 4.4044,
            "context_length": 200000,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 11
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/TitanML/GLM-5.1-FP8"
      },
      "builds": 4,
      "variants": [
        {
          "model": "GLM-5.1-RAM-270GB-MLX",
          "made_by": "baa-ai",
          "page": "/inferences/glm-5-1/base~baa-ai-ram-270gb",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "GLM-5.1-RAM-420GB-MLX",
          "made_by": "baa-ai",
          "page": "/inferences/glm-5-1/base~baa-ai-ram-420gb",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "GLM-5.1-Abliterated",
          "made_by": "helixdouble",
          "page": "/inferences/glm-5-1/base~helixdouble-abliterated",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "GLM-5.1-DQ4plus-q8",
          "made_by": "mlx-community",
          "page": "/inferences/glm-5-1/base~mlx-community-dq4plus",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "GLM-5.1-MLX-4.5bit",
          "made_by": "spicyneuron",
          "page": "/inferences/glm-5-1/base~spicyneuron-4",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not run here yet: it needs about 389.9 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $4.40 per million output tokens."
    },
    {
      "model": "DeepSeek V4.1 Flash-1017",
      "made_by": "deepseek-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/deepseek-v4-1/flash-1017",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "download_gb": 414.2,
      "params_b": 734,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v4.1-flash-1017",
        "url": "https://openrouter.ai/deepseek/deepseek-v4.1-flash-1017",
        "input_usd_per_million": 0.02,
        "output_usd_per_million": 0.108,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.02,
            "usd_per_m_out": 0.108,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.07272,
            "usd_per_m_out": 0.108,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream/flex",
            "quantization": null,
            "usd_per_m_in": 0.0775,
            "usd_per_m_out": 0.2325,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.0784,
            "usd_per_m_out": 0.2352,
            "context_length": 1048576,
            "uptime_1d": 96.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Particle AI",
            "tag": "particle-ai/us/flex",
            "quantization": null,
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.098,
            "usd_per_m_out": 0.294,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 0.0864,
            "usd_per_m_out": 0.3456,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StepFun",
            "tag": "stepfun/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.105,
            "usd_per_m_out": 0.42,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.105,
            "usd_per_m_out": 0.42,
            "context_length": 1000000,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.147,
            "usd_per_m_out": 0.441,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.111,
            "usd_per_m_out": 0.444,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.036,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Particle AI",
            "tag": "particle-ai/us",
            "quantization": null,
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.066,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.057,
            "usd_per_m_out": 0.506,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 0.55,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepSeek",
            "tag": "deepseek",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.605,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.605,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream",
            "quantization": null,
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.62,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.165,
            "usd_per_m_out": 0.66,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.68,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect",
            "quantization": null,
            "usd_per_m_in": 0.04368,
            "usd_per_m_out": 0.7,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.72,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.72,
            "context_length": 1048575,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.195,
            "usd_per_m_out": 0.78,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "OpenInference",
            "tag": "open-inference/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/spot",
            "quantization": null,
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 1048576,
            "uptime_1d": 98.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimaLabs",
            "tag": "primalabs",
            "quantization": null,
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 0.96,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.25,
            "usd_per_m_out": 1,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.16,
            "usd_per_m_out": 1.1,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 1.13,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.29,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 87.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fast",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 48
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017"
      },
      "builds": 2,
      "variants": [
        {
          "model": "DeepSeek-V4.1-Flash-1017-Uncensored-FP8",
          "made_by": "dealignai",
          "page": "/inferences/deepseek-v4-1/flash-1017~dealignai-uncensored",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "DeepSeek-V4.1-Flash-1017-GSQ-agentic-strata",
          "made_by": "kaushikvira",
          "page": "/inferences/deepseek-v4-1/flash-1017~kaushikvira-gsq-agentic-strata",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "DeepSeek-V4.1-Flash-1017-abliterated",
          "made_by": "mrfakename",
          "page": "/inferences/deepseek-v4-1/flash-1017~mrfakename-abliterated",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not run here yet: it needs about 414.2 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $0.11 per million output tokens."
    },
    {
      "model": "Muse Spark Open 816B-A42B",
      "made_by": "meta-models",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/64c115d1f661ace7c4022432/EbzfQisFJBEcKWDWLyV5O.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/muse-spark/816b-a42b",
      "hub_url": "https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B",
      "download_gb": 465.3,
      "params_b": 816,
      "active_params_b": 42,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "meta/muse-spark-open",
        "url": "https://openrouter.ai/meta/muse-spark-open",
        "input_usd_per_million": 1.25,
        "output_usd_per_million": 4.25,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 1.25,
            "usd_per_m_out": 4.25,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 1.25,
            "usd_per_m_out": 4.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.25,
            "usd_per_m_out": 4.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 1.25,
            "usd_per_m_out": 4.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.25,
            "usd_per_m_out": 4.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 5
      },
      "licence": {
        "name": "Muse Model License licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B"
      },
      "builds": 4,
      "variants": [
        {
          "model": "Muse-Spark-Open-816B-A42B-lecore-mlx",
          "made_by": "staccs",
          "page": "/inferences/muse-spark/816b-a42b~staccs-open-lecore",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not run here yet: it needs about 465.3 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $4.25 per million output tokens."
    },
    {
      "model": "DeepSeek V4.1 Flash",
      "made_by": "deepseek-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/deepseek-v4-1/flash",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash",
      "download_gb": 475.3,
      "params_b": 480,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v4.1-flash",
        "url": "https://openrouter.ai/deepseek/deepseek-v4.1-flash",
        "input_usd_per_million": 0.0507,
        "output_usd_per_million": 0.2028,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.112,
            "usd_per_m_out": 0.336,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.105,
            "usd_per_m_out": 0.42,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.42,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.48,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.135,
            "usd_per_m_out": 0.54,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.051,
            "usd_per_m_out": 0.567,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.147,
            "usd_per_m_out": 0.588,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp8",
            "usd_per_m_in": 0.103,
            "usd_per_m_out": 0.6,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.103,
            "usd_per_m_out": 0.6,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.63,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.64,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.65,
            "context_length": 1048576,
            "uptime_1d": 92.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 0.65,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.195,
            "usd_per_m_out": 0.66,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.72,
            "context_length": 1048575,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2125,
            "usd_per_m_out": 0.75,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.195,
            "usd_per_m_out": 0.78,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "OpenInference",
            "tag": "open-inference/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/spot",
            "quantization": null,
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 0.9,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.26,
            "usd_per_m_out": 1.17,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.29,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Google",
            "tag": "google-vertex/global",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/us",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/ultrafast",
            "quantization": "fp8",
            "usd_per_m_in": 3,
            "usd_per_m_out": 12,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 36
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash"
      },
      "builds": 1,
      "variants": [
        {
          "model": "DeepSeek-V4.1-Flash-1017-FP8-GGUF",
          "made_by": "AMAImedia",
          "page": "/inferences/deepseek-v4-1/flash~amaimedia-1017",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "DeepSeek-V4.1-Flash-Uncensored-FP8",
          "made_by": "dealignai",
          "page": "/inferences/deepseek-v4-1/flash~dealignai-uncensored",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not run here yet: it needs about 475.3 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $0.20 per million output tokens."
    },
    {
      "model": "MiMo V2.6 PRO-RL",
      "made_by": "XiaomiMiMo",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/680cb7d1233834890a64acee/5w_4aLfF-7MAyaIPOV498.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/mimo-v2-6/pro-rl",
      "hub_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL",
      "download_gb": 515.8,
      "params_b": 914,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "xiaomi/mimo-v2.6-pro",
        "url": "https://openrouter.ai/xiaomi/mimo-v2.6-pro",
        "input_usd_per_million": 0.435,
        "output_usd_per_million": 0.87,
        "free_tier": false,
        "context_tokens": 1050000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "GMICloud",
            "tag": "gmicloud/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.32625,
            "usd_per_m_out": 0.6525,
            "context_length": 1050000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.387,
            "usd_per_m_out": 0.783,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Xiaomi",
            "tag": "xiaomi/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.435,
            "usd_per_m_out": 0.87,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake",
            "quantization": null,
            "usd_per_m_in": 0.522,
            "usd_per_m_out": 1.044,
            "context_length": 1000000,
            "uptime_1d": 90.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimaLabs",
            "tag": "primalabs/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 4,
            "usd_per_m_out": 8,
            "context_length": 1048576,
            "uptime_1d": 97.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 5
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Pro-RL"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 515.8 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $0.87 per million output tokens."
    },
    {
      "model": "Kimi K2 Base",
      "made_by": "moonshotai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/kimi-k2/base",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K2-Instruct",
      "download_gb": 537.9,
      "params_b": 953,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k2",
        "url": "https://openrouter.ai/moonshotai/kimi-k2",
        "input_usd_per_million": 0.57,
        "output_usd_per_million": 2.3,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.57,
            "usd_per_m_out": 2.3,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K2-Instruct"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 537.9 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $2.30 per million output tokens."
    },
    {
      "model": "Kimi K2 Instruct-0905",
      "made_by": "moonshotai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/kimi-k2/instruct-0905",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K2-Instruct-0905",
      "download_gb": 537.9,
      "params_b": 953,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k2-0905",
        "url": "https://openrouter.ai/moonshotai/kimi-k2-0905",
        "input_usd_per_million": 0.6,
        "output_usd_per_million": 2.5,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K2-Instruct-0905"
      },
      "builds": 4,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 537.9 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $2.50 per million output tokens."
    },
    {
      "model": "Kimi K2.6 Base",
      "made_by": "moonshotai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/kimi-k2-6/base",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K2.6",
      "download_gb": 577.9,
      "params_b": 1024,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k2.6",
        "url": "https://openrouter.ai/moonshotai/kimi-k2.6",
        "input_usd_per_million": 0.65,
        "output_usd_per_million": 3.41,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.39,
            "usd_per_m_out": 1.8,
            "context_length": 262144,
            "uptime_1d": 98.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.57,
            "usd_per_m_out": 2.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.5985,
            "usd_per_m_out": 2.52,
            "context_length": 256000,
            "uptime_1d": 98.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/flex",
            "quantization": "fp4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.77,
            "usd_per_m_out": 3.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.8,
            "usd_per_m_out": 3.4,
            "context_length": 262144,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.65,
            "usd_per_m_out": 3.41,
            "context_length": 262144,
            "uptime_1d": 94.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 91.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 256000,
            "uptime_1d": 91.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.7985,
            "usd_per_m_out": 3.75,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 73.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/priority",
            "quantization": "fp4",
            "usd_per_m_in": 1.125,
            "usd_per_m_out": 5.25,
            "context_length": 262144,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 17
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K2.6"
      },
      "builds": 4,
      "variants": [
        {
          "model": "Kimi-K2.6-DSpark-GGUF",
          "made_by": "Anbeeld",
          "page": "/inferences/kimi-k2-6/base~anbeeld-dspark",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "Kimi-K2.6-Reap-Solidity",
          "made_by": "sombra-x",
          "page": "/inferences/kimi-k2-6/base~sombra-x-reap-solidity",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "Kimi-K2.6-MLX-3.3bit",
          "made_by": "spicyneuron",
          "page": "/inferences/kimi-k2-6/base~spicyneuron-3",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not run here yet: it needs about 577.9 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $3.41 per million output tokens."
    },
    {
      "model": "Kimi K2.7 Code",
      "made_by": "moonshotai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/kimi-k2-7/code",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K2.7-Code",
      "download_gb": 577.9,
      "params_b": 1024,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k2.7-code",
        "url": "https://openrouter.ai/moonshotai/kimi-k2.7-code",
        "input_usd_per_million": 0.7125,
        "output_usd_per_million": 3.5,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "CoreWeave",
            "tag": "coreweave/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 3,
            "context_length": 262144,
            "uptime_1d": 99.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake",
            "quantization": null,
            "usd_per_m_in": 0.7125,
            "usd_per_m_out": 3,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.85916,
            "usd_per_m_out": 3.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.912,
            "usd_per_m_out": 3.84,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai/highspeed",
            "quantization": null,
            "usd_per_m_in": 1.9,
            "usd_per_m_out": 8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 11
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K2.7-Code"
      },
      "builds": 4,
      "variants": [
        {
          "model": "Kimi-K2.7-Code-DSpark-GGUF",
          "made_by": "Anbeeld",
          "page": "/inferences/kimi-k2-7/code~anbeeld-dspark",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "Kimi-K2.7-Code-MLX-4bit-hiprec",
          "made_by": "pipenetwork",
          "page": "/inferences/kimi-k2-7/code~pipenetwork-hiprec",
          "status": "not measured yet",
          "tokens_per_sec": null
        },
        {
          "model": "Kimi-K2.7-Code-MLX-3.6bit",
          "made_by": "spicyneuron",
          "page": "/inferences/kimi-k2-7/code~spicyneuron-3",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not run here yet: it needs about 577.9 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $3.50 per million output tokens."
    },
    {
      "model": "Kimi K2 Thinking",
      "made_by": "moonshotai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/kimi-k2/thinking",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K2-Thinking",
      "download_gb": 578.6,
      "params_b": 1025,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k2-thinking",
        "url": "https://openrouter.ai/moonshotai/kimi-k2-thinking",
        "input_usd_per_million": 0.6,
        "output_usd_per_million": 2.5,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Novita",
            "tag": "novita/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K2-Thinking"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 578.6 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $2.50 per million output tokens."
    },
    {
      "model": "Kimi K2.5 Base",
      "made_by": "moonshotai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/kimi-k2-5/base",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K2.5",
      "download_gb": 578.6,
      "params_b": 1025,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k2.5",
        "url": "https://openrouter.ai/moonshotai/kimi-k2.5",
        "input_usd_per_million": 0.5,
        "output_usd_per_million": 2.5,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.49,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.5,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.57,
            "usd_per_m_out": 2.85,
            "context_length": 262144,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock/us-east-2",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 3,
            "context_length": 262144,
            "uptime_1d": 98.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 0.532,
            "usd_per_m_out": 3.325,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 5
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K2.5"
      },
      "builds": 3,
      "variants": [
        {
          "model": "Kimi-K2.5-MLX-2.8bit",
          "made_by": "spicyneuron",
          "page": "/inferences/kimi-k2-5/base~spicyneuron-2",
          "status": "not measured yet",
          "tokens_per_sec": null
        }
      ],
      "bottom_line": "Not run here yet: it needs about 578.6 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $2.50 per million output tokens."
    },
    {
      "model": "MiMo V2.5 PRO",
      "made_by": "XiaomiMiMo",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/680cb7d1233834890a64acee/5w_4aLfF-7MAyaIPOV498.jpeg",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/mimo-v2-5/pro",
      "hub_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.5-Pro",
      "download_gb": 586.4,
      "params_b": 1039,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "xiaomi/mimo-v2.5-pro",
        "url": "https://openrouter.ai/xiaomi/mimo-v2.5-pro",
        "input_usd_per_million": 0.435,
        "output_usd_per_million": 0.87,
        "free_tier": false,
        "context_tokens": 1024000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.435,
            "usd_per_m_out": 0.87,
            "context_length": 1024000,
            "uptime_1d": 98.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.48,
            "usd_per_m_out": 1.8,
            "context_length": 262144,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.5-Pro"
      },
      "builds": 4,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 586.4 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $0.87 per million output tokens."
    },
    {
      "model": "DeepSeek V3.2 EXP",
      "made_by": "deepseek-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/deepseek-v3-2/exp",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V3.2-Exp",
      "download_gb": 642.1,
      "params_b": 649,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v3.2-exp",
        "url": "https://openrouter.ai/deepseek/deepseek-v3.2-exp",
        "input_usd_per_million": 0.27,
        "output_usd_per_million": 0.41,
        "free_tier": false,
        "context_tokens": 163840,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.27,
            "usd_per_m_out": 0.41,
            "context_length": 163840,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V3.2-Exp"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 642.1 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $0.41 per million output tokens."
    },
    {
      "model": "GLM-5.3 Base",
      "made_by": "zai-org",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/glm-5-3/base",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.3",
      "download_gb": 703.7,
      "params_b": 711,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.3",
        "url": "https://openrouter.ai/z-ai/glm-5.3",
        "input_usd_per_million": 0.084,
        "output_usd_per_million": 3.6,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.0699,
            "usd_per_m_out": 1.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.179,
            "usd_per_m_out": 1.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Decart",
            "tag": "decart/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.56,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.63,
            "usd_per_m_out": 1.98,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/flex",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Tenstorrent",
            "tag": "tenstorrent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2,
            "context_length": 1044480,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 2.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 2.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 2.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.5625,
            "usd_per_m_out": 2.5,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.91,
            "usd_per_m_out": 2.86,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 3,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3,
            "context_length": 1048576,
            "uptime_1d": 95,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/us",
            "quantization": "fp8",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/spot",
            "quantization": null,
            "usd_per_m_in": 0.98,
            "usd_per_m_out": 3.08,
            "context_length": 1048576,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 3.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.084,
            "usd_per_m_out": 3.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.084,
            "usd_per_m_out": 3.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 1.19,
            "usd_per_m_out": 3.74,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.275,
            "usd_per_m_out": 3.921,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 1.26,
            "usd_per_m_out": 3.96,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1,
            "usd_per_m_out": 4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1,
            "usd_per_m_out": 4.2,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.05,
            "usd_per_m_out": 4.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.0703,
            "usd_per_m_out": 4.3843,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Doubleword",
            "tag": "doubleword",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Featherless",
            "tag": "featherless",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 262144,
            "uptime_1d": 97,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SambaNova",
            "tag": "sambanova",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": "nvfp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp4/us",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 84.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/nvfp4/eu",
            "quantization": "nvfp4",
            "usd_per_m_in": 1.54,
            "usd_per_m_out": 4.84,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock/us-east-1",
            "quantization": null,
            "usd_per_m_in": 1.68,
            "usd_per_m_out": 5.28,
            "context_length": 1048576,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp4",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 96.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/fast",
            "quantization": null,
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/us/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/prime",
            "quantization": null,
            "usd_per_m_in": 2.8,
            "usd_per_m_out": 8.8,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 49
      },
      "licence": {
        "name": "GLM 5.3 licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/zai-org/GLM-5.3"
      },
      "builds": 3,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 703.7 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $3.60 per million output tokens."
    },
    {
      "model": "DeepSeek V4 PRO",
      "made_by": "deepseek-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/deepseek-v4/pro",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro",
      "download_gb": 805.3,
      "params_b": 814,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v4-pro",
        "url": "https://openrouter.ai/deepseek/deepseek-v4-pro",
        "input_usd_per_million": 0.9,
        "output_usd_per_million": 1.5,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.9,
            "usd_per_m_out": 1.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1,
            "usd_per_m_out": 1.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.9483,
            "usd_per_m_out": 1.8966,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.957,
            "usd_per_m_out": 1.914,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 1.04,
            "usd_per_m_out": 2.08,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 1.044,
            "usd_per_m_out": 2.088,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.3,
            "usd_per_m_out": 2.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.416,
            "usd_per_m_out": 2.832,
            "context_length": 1000000,
            "uptime_1d": 96.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8/us-east-1",
            "quantization": "fp8",
            "usd_per_m_in": 1.56,
            "usd_per_m_out": 3.12,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.6,
            "usd_per_m_out": 3.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 1.65,
            "usd_per_m_out": 3.301,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.68,
            "usd_per_m_out": 3.38,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.69,
            "usd_per_m_out": 3.38,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Azure",
            "tag": "azure/us",
            "quantization": null,
            "usd_per_m_in": 1.91,
            "usd_per_m_out": 3.83,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 1.95,
            "usd_per_m_out": 3.9,
            "context_length": 1048576,
            "uptime_1d": 99.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 15
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 805.3 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $1.50 per million output tokens."
    },
    {
      "model": "DeepSeek V4 PRO-0813",
      "made_by": "deepseek-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": false,
      "page": "/inferences/deepseek-v4/pro-0813",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813",
      "download_gb": 831.4,
      "params_b": 840,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v4-pro-0813",
        "url": "https://openrouter.ai/deepseek/deepseek-v4-pro-0813",
        "input_usd_per_million": 0.111,
        "output_usd_per_million": 4.44,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.5808,
            "usd_per_m_out": 1.7424,
            "context_length": 1000000,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/us-east-1",
            "quantization": null,
            "usd_per_m_in": 0.66,
            "usd_per_m_out": 1.98,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake",
            "quantization": null,
            "usd_per_m_in": 0.66,
            "usd_per_m_out": 1.98,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepSeek",
            "tag": "deepseek",
            "quantization": null,
            "usd_per_m_in": 0.66,
            "usd_per_m_out": 1.98,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 1.04,
            "usd_per_m_out": 2.08,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/us",
            "quantization": "fp4",
            "usd_per_m_in": 1.35,
            "usd_per_m_out": 2.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.35,
            "usd_per_m_out": 2.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.3,
            "usd_per_m_out": 2.6,
            "context_length": 1048576,
            "uptime_1d": 98.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.99,
            "usd_per_m_out": 2.97,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.056,
            "usd_per_m_out": 3.168,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 1.95,
            "usd_per_m_out": 3.9,
            "context_length": 1048576,
            "uptime_1d": 97.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.31,
            "usd_per_m_out": 3.96,
            "context_length": 1048576,
            "uptime_1d": 96.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.32,
            "usd_per_m_out": 3.96,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.32,
            "usd_per_m_out": 3.96,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.32,
            "usd_per_m_out": 3.96,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.32,
            "usd_per_m_out": 3.96,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.32,
            "usd_per_m_out": 3.96,
            "context_length": 1048576,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 1.32,
            "usd_per_m_out": 3.96,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 1.32,
            "usd_per_m_out": 3.96,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.32,
            "usd_per_m_out": 3.96,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 1.32,
            "usd_per_m_out": 3.96,
            "context_length": 1048576,
            "uptime_1d": 89.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.285,
            "usd_per_m_out": 4.3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.111,
            "usd_per_m_out": 4.44,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 1.65,
            "usd_per_m_out": 4.95,
            "context_length": 1000000,
            "uptime_1d": 98,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp8",
            "usd_per_m_in": 1.98,
            "usd_per_m_out": 5.94,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.0999,
            "usd_per_m_out": 8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 26
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Pro-0813"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 831.4 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $4.44 per million output tokens."
    },
    {
      "model": "hy3-tiny-random",
      "made_by": "yujiepan",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/Lp3m-XLpjQGwBItlvn69q.png",
      "based_on": "Hunyuan Hy3 Base",
      "variant_of": "/inferences/hy-3/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/hy-3/base~yujiepan-tiny-random",
      "hub_url": "https://huggingface.co/tencent/Hy3",
      "download_gb": 0,
      "params_b": null,
      "active_params_b": null,
      "params_basis": null,
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [
        "2026-W45"
      ],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "tencent/hy3",
        "url": "https://openrouter.ai/tencent/hy3",
        "input_usd_per_million": 0.0825,
        "output_usd_per_million": 0.33,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Tencent",
            "tag": "tencent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.0825,
            "usd_per_m_out": 0.33,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.58,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.58,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 0.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 4
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/tencent/Hy3"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.33 per million output tokens."
    },
    {
      "model": "MiniMax-M3-DSpark-GGUF",
      "made_by": "Anbeeld",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/676e38ad04af5bec20bc9faf/dUd-LsZEX0H_d4qefO_g6.jpeg",
      "based_on": "MiniMax-M3 Base",
      "variant_of": "/inferences/minimax-m3/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/minimax-m3/base~anbeeld-dspark",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M3",
      "download_gb": 1.1,
      "params_b": 2,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m3",
        "url": "https://openrouter.ai/minimax/minimax-m3",
        "input_usd_per_million": 0.3,
        "output_usd_per_million": 1.2,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.224,
            "usd_per_m_out": 0.88,
            "context_length": 524288,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.23,
            "usd_per_m_out": 0.96,
            "context_length": 262144,
            "uptime_1d": 91.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 0.96,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 1.1,
            "context_length": 524288,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 524288,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Minimax",
            "tag": "minimax",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 524288,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 8
      },
      "licence": {
        "name": "Minimax Community licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/MiniMaxAI/MiniMax-M3"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.20 per million output tokens."
    },
    {
      "model": "Kimi-K2.6-DSpark-GGUF",
      "made_by": "Anbeeld",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": "Kimi K2.6 Base",
      "variant_of": "/inferences/kimi-k2-6/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/kimi-k2-6/base~anbeeld-dspark",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K2.6",
      "download_gb": 1.7,
      "params_b": 3.1,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k2.6",
        "url": "https://openrouter.ai/moonshotai/kimi-k2.6",
        "input_usd_per_million": 0.65,
        "output_usd_per_million": 3.41,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.39,
            "usd_per_m_out": 1.8,
            "context_length": 262144,
            "uptime_1d": 98.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.57,
            "usd_per_m_out": 2.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.5985,
            "usd_per_m_out": 2.52,
            "context_length": 256000,
            "uptime_1d": 98.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/flex",
            "quantization": "fp4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.77,
            "usd_per_m_out": 3.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.8,
            "usd_per_m_out": 3.4,
            "context_length": 262144,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.65,
            "usd_per_m_out": 3.41,
            "context_length": 262144,
            "uptime_1d": 94.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 91.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 256000,
            "uptime_1d": 91.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.7985,
            "usd_per_m_out": 3.75,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 73.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/priority",
            "quantization": "fp4",
            "usd_per_m_in": 1.125,
            "usd_per_m_out": 5.25,
            "context_length": 262144,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 17
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K2.6"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $3.41 per million output tokens."
    },
    {
      "model": "Kimi-K2.7-Code-DSpark-GGUF",
      "made_by": "Anbeeld",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": "Kimi K2.7 Code",
      "variant_of": "/inferences/kimi-k2-7/code",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/kimi-k2-7/code~anbeeld-dspark",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K2.7-Code",
      "download_gb": 1.7,
      "params_b": 3.1,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k2.7-code",
        "url": "https://openrouter.ai/moonshotai/kimi-k2.7-code",
        "input_usd_per_million": 0.7125,
        "output_usd_per_million": 3.5,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "CoreWeave",
            "tag": "coreweave/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 3,
            "context_length": 262144,
            "uptime_1d": 99.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake",
            "quantization": null,
            "usd_per_m_in": 0.7125,
            "usd_per_m_out": 3,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.85916,
            "usd_per_m_out": 3.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.912,
            "usd_per_m_out": 3.84,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai/highspeed",
            "quantization": null,
            "usd_per_m_in": 1.9,
            "usd_per_m_out": 8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 11
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K2.7-Code"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $3.50 per million output tokens."
    },
    {
      "model": "Llama-3.2-3B-Instruct-abliterated-GGUF",
      "made_by": "QuantFactory",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6382255fcae34727b9cc149e/ANRA_7hWosC6_2PS2cwtg.jpeg",
      "based_on": "Llama 3.2 3B",
      "variant_of": "/inferences/llama-3-2/3b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/llama-3-2/3b~quantfactory-llama-3-2-abliterated",
      "hub_url": "https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct",
      "download_gb": 1.9,
      "params_b": 3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "meta-llama/llama-3.2-3b-instruct",
        "url": "https://openrouter.ai/meta-llama/llama-3.2-3b-instruct",
        "input_usd_per_million": 0.0509,
        "output_usd_per_million": 0.335,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Parasail",
            "tag": "parasail/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.05,
            "usd_per_m_out": 0.33,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.0509,
            "usd_per_m_out": 0.335,
            "context_length": 80000,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Llama 3.2 community licence",
        "you_may_serve": "Anyone, under the maker's terms",
        "class": "open-weights",
        "source": "https://huggingface.co/meta-llama/Llama-3.2-3B-Instruct"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.34 per million output tokens."
    },
    {
      "model": "Qwen3.5-9B",
      "made_by": "unsloth",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62ecdc18b72a69615d6bd857/E4lkPz1TZNLzIFr_dR273.png",
      "based_on": "DeepSeek V4",
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/deepseek-v4/9b~unsloth-qwen3-5",
      "hub_url": "https://huggingface.co/unsloth/Qwen3.5-9B",
      "download_gb": 5.2,
      "params_b": 9,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.5-9b",
        "url": "https://openrouter.ai/qwen/qwen3.5-9b",
        "input_usd_per_million": 0.037,
        "output_usd_per_million": 0.054,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16/flex",
            "quantization": "bf16",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.12,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.13,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.15,
            "context_length": 262144,
            "uptime_1d": 57.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.15,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.148,
            "usd_per_m_out": 0.218,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16/priority",
            "quantization": "bf16",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.225,
            "context_length": 262144,
            "uptime_1d": 96.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.25,
            "context_length": 262144,
            "uptime_1d": 60.4,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.17,
            "usd_per_m_out": 0.25,
            "context_length": 262144,
            "uptime_1d": 97.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 8
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/unsloth/Qwen3.5-9B"
      },
      "builds": 2,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.05 per million output tokens."
    },
    {
      "model": "Qwen3.5-9B-Artemis-GGUF",
      "made_by": "turintech",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": "Qwen 3.5 9B",
      "variant_of": "/inferences/qwen-3-5/9b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/qwen-3-5/9b~turintech-artemis",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.5-9B",
      "download_gb": 5.3,
      "params_b": 9,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.5-9b",
        "url": "https://openrouter.ai/qwen/qwen3.5-9b",
        "input_usd_per_million": 0.037,
        "output_usd_per_million": 0.054,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16/flex",
            "quantization": "bf16",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.12,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.13,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.15,
            "context_length": 262144,
            "uptime_1d": 57.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.15,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.148,
            "usd_per_m_out": 0.218,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16/priority",
            "quantization": "bf16",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.225,
            "context_length": 262144,
            "uptime_1d": 96.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.25,
            "context_length": 262144,
            "uptime_1d": 60.4,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.17,
            "usd_per_m_out": 0.25,
            "context_length": 262144,
            "uptime_1d": 97.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 8
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.5-9B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.05 per million output tokens."
    },
    {
      "model": "LFM2.5-24B-A2B-Uncensored-GGUF",
      "made_by": "AtomicChat",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/61b8e2ba285851687028d395/EsTgVtnM2IqVRKgPdfqcB.png",
      "based_on": "LFM2.5 24B-A2B",
      "variant_of": "/inferences/lfm-2-5/24b-a2b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/lfm-2-5/24b-a2b~atomicchat-uncensored",
      "hub_url": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B",
      "download_gb": 14.3,
      "params_b": 24,
      "active_params_b": 2,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "liquid/lfm-2.5-24b-a2b",
        "url": "https://openrouter.ai/liquid/lfm-2.5-24b-a2b",
        "input_usd_per_million": 0.05,
        "output_usd_per_million": 0.7,
        "free_tier": false,
        "context_tokens": 128000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Liquid",
            "tag": "liquid/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.05,
            "usd_per_m_out": 0.7,
            "context_length": 128000,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "LFM1.0 licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/LiquidAI/LFM2.5-24B-A2B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.70 per million output tokens."
    },
    {
      "model": "AX-Hy-MT2-30B-A3B-MLX-AXQ-MXFP4",
      "made_by": "AutomatosX",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/Lp3m-XLpjQGwBItlvn69q.png",
      "based_on": "Hunyuan MT 2 30B-A3B",
      "variant_of": "/inferences/hy-mt-2/30b-a3b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/hy-mt-2/30b-a3b~automatosx-ax-axq",
      "hub_url": "https://huggingface.co/tencent/Hy-MT2-30B-A3B",
      "download_gb": 15.1,
      "params_b": 30,
      "active_params_b": 3,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "tencent/hy-mt2-30b-a3b",
        "url": "https://openrouter.ai/tencent/hy-mt2-30b-a3b",
        "input_usd_per_million": 0.074,
        "output_usd_per_million": 0.295,
        "free_tier": false,
        "context_tokens": 8192,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Tencent",
            "tag": "tencent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.074,
            "usd_per_m_out": 0.295,
            "context_length": 8192,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/tencent/Hy-MT2-30B-A3B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.29 per million output tokens."
    },
    {
      "model": "Muse-Glimmer-30B-pruned-V2.5-en-pl-GGUF",
      "made_by": "bluevoid-pl",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/64c115d1f661ace7c4022432/EbzfQisFJBEcKWDWLyV5O.png",
      "based_on": "Muse Glimmer 30B",
      "variant_of": "/inferences/muse-glimmer/30b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/muse-glimmer/30b~bluevoid-pl-pruned-v2-5-en",
      "hub_url": "https://huggingface.co/meta-models/Muse-Glimmer-30B",
      "download_gb": 15.2,
      "params_b": 30,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "meta/muse-glimmer-30b",
        "url": "https://openrouter.ai/meta/muse-glimmer-30b",
        "input_usd_per_million": 0.3,
        "output_usd_per_million": 1.2,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.35,
            "usd_per_m_out": 1.5,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/meta-models/Muse-Glimmer-30B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.20 per million output tokens."
    },
    {
      "model": "Huihui-GLM-4.7-Flash-abliterated-mlx-4bit",
      "made_by": "huihui-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/66228dc4c8920ec3513dc81a/9stczijVqp6TgQduzT8kN.png",
      "based_on": "GLM-4.7 Flash",
      "variant_of": "/inferences/glm-4-7/flash",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-4-7/flash~huihui-ai-abliterated",
      "hub_url": "https://huggingface.co/zai-org/GLM-4.7-Flash",
      "download_gb": 15.7,
      "params_b": 28,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-4.7-flash",
        "url": "https://openrouter.ai/z-ai/glm-4.7-flash",
        "input_usd_per_million": 0.0605,
        "output_usd_per_million": 0.4,
        "free_tier": false,
        "context_tokens": 200000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.0605,
            "usd_per_m_out": 0.4,
            "context_length": 131072,
            "uptime_1d": 99.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.4,
            "context_length": 200000,
            "uptime_1d": 5.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-4.7-Flash"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.40 per million output tokens."
    },
    {
      "model": "Muse-Glimmer-30B-Heretic-Uncensored-GGUF",
      "made_by": "0bserverx",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/64c115d1f661ace7c4022432/EbzfQisFJBEcKWDWLyV5O.png",
      "based_on": "Muse Glimmer 30B",
      "variant_of": "/inferences/muse-glimmer/30b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/muse-glimmer/30b~0bserverx-heretic-uncensored",
      "hub_url": "https://huggingface.co/meta-models/Muse-Glimmer-30B",
      "download_gb": 15.8,
      "params_b": 30,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "meta/muse-glimmer-30b",
        "url": "https://openrouter.ai/meta/muse-glimmer-30b",
        "input_usd_per_million": 0.3,
        "output_usd_per_million": 1.2,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.35,
            "usd_per_m_out": 1.5,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/meta-models/Muse-Glimmer-30B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.20 per million output tokens."
    },
    {
      "model": "GLM-4.7-Flash-RAM-20GB-MLX",
      "made_by": "baa-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/644d2651d185572dd1e497d9/-F6ZUGFzgXLlFa1asCLBW.png",
      "based_on": "GLM-4.7 Flash",
      "variant_of": "/inferences/glm-4-7/flash",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-4-7/flash~baa-ai-ram-20gb",
      "hub_url": "https://huggingface.co/zai-org/GLM-4.7-Flash",
      "download_gb": 17.9,
      "params_b": 32,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-4.7-flash",
        "url": "https://openrouter.ai/z-ai/glm-4.7-flash",
        "input_usd_per_million": 0.0605,
        "output_usd_per_million": 0.4,
        "free_tier": false,
        "context_tokens": 200000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.0605,
            "usd_per_m_out": 0.4,
            "context_length": 131072,
            "uptime_1d": 99.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.4,
            "context_length": 200000,
            "uptime_1d": 5.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-4.7-Flash"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.40 per million output tokens."
    },
    {
      "model": "Qwen3.6-35B-A3B",
      "made_by": "Qwen",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": "Hunyuan Hy3",
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/hy-3/35b-a3b~qwen-qwen3-6",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B",
      "download_gb": 18.6,
      "params_b": 35,
      "active_params_b": 3,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.6-35b-a3b",
        "url": "https://openrouter.ai/qwen/qwen3.6-35b-a3b",
        "input_usd_per_million": 0.1,
        "output_usd_per_million": 0.9,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.066,
            "usd_per_m_out": 0.602,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.05,
            "usd_per_m_out": 0.7,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.76,
            "context_length": 262144,
            "uptime_1d": 85.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.89,
            "context_length": 262144,
            "uptime_1d": 91.9,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.9,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.9,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.95,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 1,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 1,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.25,
            "usd_per_m_out": 1.25,
            "context_length": 262144,
            "uptime_1d": 29.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.425,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 1.8,
            "context_length": 262144,
            "uptime_1d": 95.3,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 12
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.90 per million output tokens."
    },
    {
      "model": "Muse-Glimmer-30B-VL-gptq-mlx-jang",
      "made_by": "roman220220",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/64c115d1f661ace7c4022432/EbzfQisFJBEcKWDWLyV5O.png",
      "based_on": "Muse Glimmer 30B",
      "variant_of": "/inferences/muse-glimmer/30b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/muse-glimmer/30b~roman220220-vl-jang",
      "hub_url": "https://huggingface.co/meta-models/Muse-Glimmer-30B",
      "download_gb": 20,
      "params_b": 30,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "meta/muse-glimmer-30b",
        "url": "https://openrouter.ai/meta/muse-glimmer-30b",
        "input_usd_per_million": 0.3,
        "output_usd_per_million": 1.2,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.35,
            "usd_per_m_out": 1.5,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/meta-models/Muse-Glimmer-30B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.20 per million output tokens."
    },
    {
      "model": "Qwen3.6-35B-A3B-BE-GGUF",
      "made_by": "taronaeo",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": "Qwen 3.6 35B-A3B",
      "variant_of": "/inferences/qwen-3-6/35b-a3b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/qwen-3-6/35b-a3b~taronaeo-be",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B",
      "download_gb": 20.2,
      "params_b": 35,
      "active_params_b": 3,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.6-35b-a3b",
        "url": "https://openrouter.ai/qwen/qwen3.6-35b-a3b",
        "input_usd_per_million": 0.1,
        "output_usd_per_million": 0.9,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.066,
            "usd_per_m_out": 0.602,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.05,
            "usd_per_m_out": 0.7,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.76,
            "context_length": 262144,
            "uptime_1d": 85.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.89,
            "context_length": 262144,
            "uptime_1d": 91.9,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.9,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.9,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.95,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 1,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 1,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.25,
            "usd_per_m_out": 1.25,
            "context_length": 262144,
            "uptime_1d": 29.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.425,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 1.8,
            "context_length": 262144,
            "uptime_1d": 95.3,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 12
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.90 per million output tokens."
    },
    {
      "model": "Mistral-Small-4-119B-2603-Mixed-3bit-MLX",
      "made_by": "exalandru",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/634c17653d11eaedd88b314d/9OgyfKstSZtbmsmuG8MbU.png",
      "based_on": "Mistral Small 4 119B-2603",
      "variant_of": "/inferences/mistral-small-4/119b-2603",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/mistral-small-4/119b-2603~exalandru-mixed",
      "hub_url": "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603",
      "download_gb": 45,
      "params_b": 119,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "mistralai/mistral-small-2603",
        "url": "https://openrouter.ai/mistralai/mistral-small-2603",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 0.6,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 0.165,
            "usd_per_m_out": 0.66,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/us",
            "quantization": null,
            "usd_per_m_in": 0.165,
            "usd_per_m_out": 0.66,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 3
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.60 per million output tokens."
    },
    {
      "model": "Huihui-GLM-4.5V-abliterated-mxfp4",
      "made_by": "mlx-community",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/623c830997ddced06d78699b/3qTjC7d3YFCJTwpxd2noq.png",
      "based_on": "GLM-4.5",
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-4-5/v~mlx-community-huihui-abliterated",
      "hub_url": "https://huggingface.co/zai-org/GLM-4.5V",
      "download_gb": 54.5,
      "params_b": 97,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-4.5v",
        "url": "https://openrouter.ai/z-ai/glm-4.5v",
        "input_usd_per_million": 0.6,
        "output_usd_per_million": 1.8,
        "free_tier": false,
        "context_tokens": 65536,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.8,
            "context_length": 65536,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.8,
            "context_length": 65536,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-4.5V"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.80 per million output tokens."
    },
    {
      "model": "GLM-4.5V-4bit",
      "made_by": "mlx-community",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/623c830997ddced06d78699b/3qTjC7d3YFCJTwpxd2noq.png",
      "based_on": "GLM-4.5",
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-4-5/v~mlx-community",
      "hub_url": "https://huggingface.co/mlx-community/GLM-4.5V-4bit",
      "download_gb": 57.6,
      "params_b": 102,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-4.5v",
        "url": "https://openrouter.ai/z-ai/glm-4.5v",
        "input_usd_per_million": 0.6,
        "output_usd_per_million": 1.8,
        "free_tier": false,
        "context_tokens": 65536,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.8,
            "context_length": 65536,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.8,
            "context_length": 65536,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/mlx-community/GLM-4.5V-4bit"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.80 per million output tokens."
    },
    {
      "model": "GLM-4.5-Air-mixed_4_6",
      "made_by": "terion-mlx",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": "GLM-4.5 AIR",
      "variant_of": "/inferences/glm-4-5/air",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-4-5/air~terion-mlx-mixed-4-6",
      "hub_url": "https://huggingface.co/zai-org/GLM-4.5-Air",
      "download_gb": 59.8,
      "params_b": 106,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-4.5-air",
        "url": "https://openrouter.ai/z-ai/glm-4.5-air",
        "input_usd_per_million": 0.2,
        "output_usd_per_million": 1.1,
        "free_tier": false,
        "context_tokens": 131072,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Novita",
            "tag": "novita/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.85,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.86,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 1.1,
            "context_length": 131072,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 3
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-4.5-Air"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.10 per million output tokens."
    },
    {
      "model": "Mistral-Small-4-119B-TurboQuant-MLX-4bit",
      "made_by": "majentik",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/634c17653d11eaedd88b314d/9OgyfKstSZtbmsmuG8MbU.png",
      "based_on": "Mistral Small 4 119B-2603",
      "variant_of": "/inferences/mistral-small-4/119b-2603",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/mistral-small-4/119b-2603~majentik-turboquant",
      "hub_url": "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603",
      "download_gb": 63.1,
      "params_b": 119,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "mistralai/mistral-small-2603",
        "url": "https://openrouter.ai/mistralai/mistral-small-2603",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 0.6,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 0.165,
            "usd_per_m_out": 0.66,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/us",
            "quantization": null,
            "usd_per_m_in": 0.165,
            "usd_per_m_out": 0.66,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 3
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/mistralai/Mistral-Small-4-119B-2603"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.60 per million output tokens."
    },
    {
      "model": "GLM-4.5V-i1-GGUF",
      "made_by": "mradermacher",
      "maker_logo": "https://huggingface.co/avatars/96d0c12e8eb411b44ecefc1b6508a420.svg",
      "based_on": "GLM-4.5",
      "variant_of": null,
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-4-5/v~mradermacher-i1",
      "hub_url": "https://huggingface.co/zai-org/GLM-4.5V",
      "download_gb": 65.6,
      "params_b": 116,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-4.5v",
        "url": "https://openrouter.ai/z-ai/glm-4.5v",
        "input_usd_per_million": 0.6,
        "output_usd_per_million": 1.8,
        "free_tier": false,
        "context_tokens": 65536,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.8,
            "context_length": 65536,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.8,
            "context_length": 65536,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-4.5V"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.80 per million output tokens."
    },
    {
      "model": "Qwen3.8-Flash-Next-Reap-288-MLX-4bit",
      "made_by": "sh0wie",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6215ca5692c0ecfba9186921/hrRM50-6XcdWgg2AKpENG.jpeg",
      "based_on": "Qwen 3.8 Flash-Next",
      "variant_of": "/inferences/qwen-3-8/flash-next",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/qwen-3-8/flash-next~sh0wie-reap-288",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next",
      "download_gb": 68.4,
      "params_b": 121,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.8-flash",
        "url": "https://openrouter.ai/qwen/qwen3.8-flash",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 0.47,
        "free_tier": false,
        "context_tokens": 1000000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Alibaba",
            "tag": "alibaba/us-east-1",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.47,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.47,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Qwen Community 1.0 licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.47 per million output tokens."
    },
    {
      "model": "Mistral-Medium-3.5-128B-Eschaton-Uncensored-GGUF",
      "made_by": "cloudbjorn",
      "maker_logo": "https://huggingface.co/avatars/3c46602429746b1bebbbb7e57f8d64ee.svg",
      "based_on": "Mistral Medium 3.5 128B",
      "variant_of": "/inferences/mistral-medium-3-5/128b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/mistral-medium-3-5/128b~cloudbjorn-eschaton-uncensored",
      "hub_url": "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B",
      "download_gb": 69.8,
      "params_b": 128,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "mistralai/mistral-medium-3-5",
        "url": "https://openrouter.ai/mistralai/mistral-medium-3-5",
        "input_usd_per_million": 1.5,
        "output_usd_per_million": 7.5,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": null,
            "usd_per_m_in": 1.5,
            "usd_per_m_out": 7.5,
            "context_length": 262144,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 1.65,
            "usd_per_m_out": 8.25,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $7.50 per million output tokens."
    },
    {
      "model": "MiniMax-M2.7-ConfigI-MLX",
      "made_by": "thetom-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/676e38ad04af5bec20bc9faf/dUd-LsZEX0H_d4qefO_g6.jpeg",
      "based_on": "MiniMax-M2.7 Base",
      "variant_of": "/inferences/minimax-m2-7/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/minimax-m2-7/base~thetom-ai-configi",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7",
      "download_gb": 86.5,
      "params_b": 153,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m2.7",
        "url": "https://openrouter.ai/minimax/minimax-m2.7",
        "input_usd_per_million": 0.21,
        "output_usd_per_million": 0.84,
        "free_tier": false,
        "context_tokens": 196608,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 196608,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Groq",
            "tag": "groq",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.8,
            "context_length": 196608,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.84 per million output tokens."
    },
    {
      "model": "Qwen3.8-27B-Uncensored-MLX",
      "made_by": "orcarouter",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/65224256377bffdc599901c9/-qy82WgzPEAjof3QPGftY.png",
      "based_on": "Qwen 3.8 27B",
      "variant_of": "/inferences/qwen-3-8/27b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/qwen-3-8/27b~orcarouter-uncensored",
      "hub_url": "https://huggingface.co/Qwen/Qwen3.8-27B",
      "download_gb": 88.1,
      "params_b": 27,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "qwen/qwen3.8-27b",
        "url": "https://openrouter.ai/qwen/qwen3.8-27b",
        "input_usd_per_million": 0.15,
        "output_usd_per_million": 1.875,
        "free_tier": false,
        "context_tokens": 1024000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.11,
            "usd_per_m_out": 0.781,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake",
            "quantization": null,
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 1.05,
            "context_length": 1024000,
            "uptime_1d": 97.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.039,
            "usd_per_m_out": 1.25,
            "context_length": 262144,
            "uptime_1d": 91.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 1.35,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cerebras",
            "tag": "cerebras/fp16",
            "quantization": "fp16",
            "usd_per_m_in": 0.99,
            "usd_per_m_out": 1.49,
            "context_length": 65536,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.035,
            "usd_per_m_out": 1.5,
            "context_length": 262144,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Tenstorrent",
            "tag": "tenstorrent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.6,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.875,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 1.95,
            "context_length": 262144,
            "uptime_1d": 91.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AkashML",
            "tag": "akashml/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 1.98,
            "context_length": 262144,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.125,
            "usd_per_m_out": 2,
            "context_length": 262144,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 2.2,
            "context_length": 262144,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 2.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 2.3,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.04,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.425,
            "usd_per_m_out": 2.55,
            "context_length": 1000000,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.4,
            "usd_per_m_out": 3,
            "context_length": 262144,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.42,
            "usd_per_m_out": 3,
            "context_length": 1000000,
            "uptime_1d": 79.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 95.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 3.2,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 23
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/Qwen/Qwen3.8-27B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.88 per million output tokens."
    },
    {
      "model": "MiniMax-M2.7-TQ3",
      "made_by": "fno2010",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/676e38ad04af5bec20bc9faf/dUd-LsZEX0H_d4qefO_g6.jpeg",
      "based_on": "MiniMax-M2.7 Base",
      "variant_of": "/inferences/minimax-m2-7/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/minimax-m2-7/base~fno2010-tq3",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7",
      "download_gb": 88.4,
      "params_b": 89,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m2.7",
        "url": "https://openrouter.ai/minimax/minimax-m2.7",
        "input_usd_per_million": 0.21,
        "output_usd_per_million": 0.84,
        "free_tier": false,
        "context_tokens": 196608,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 196608,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Groq",
            "tag": "groq",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.8,
            "context_length": 196608,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.84 per million output tokens."
    },
    {
      "model": "MiniMax-M2.7-RAM-91GB-MLX",
      "made_by": "baa-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/644d2651d185572dd1e497d9/-F6ZUGFzgXLlFa1asCLBW.png",
      "based_on": "MiniMax-M2.7 Base",
      "variant_of": "/inferences/minimax-m2-7/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/minimax-m2-7/base~baa-ai-ram-91gb",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7",
      "download_gb": 89.8,
      "params_b": 159,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m2.7",
        "url": "https://openrouter.ai/minimax/minimax-m2.7",
        "input_usd_per_million": 0.21,
        "output_usd_per_million": 0.84,
        "free_tier": false,
        "context_tokens": 196608,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 196608,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Groq",
            "tag": "groq",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.8,
            "context_length": 196608,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.84 per million output tokens."
    },
    {
      "model": "GLM-5.3-Flash-REAP50-GGUF",
      "made_by": "patrickbdevaney",
      "maker_logo": "https://huggingface.co/avatars/f24ba3a80e566218af6dba046b5b4b90.svg",
      "based_on": "GLM-5.3 Flash",
      "variant_of": "/inferences/glm-5-3/flash",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-5-3/flash~patrickbdevaney-reap50",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.3-Flash",
      "download_gb": 92.5,
      "params_b": 164,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.3-flash",
        "url": "https://openrouter.ai/z-ai/glm-5.3-flash",
        "input_usd_per_million": 0.0399,
        "output_usd_per_million": 0.0784,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.0399,
            "usd_per_m_out": 0.0784,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.065,
            "usd_per_m_out": 0.18,
            "context_length": 262144,
            "uptime_1d": 96.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.174,
            "usd_per_m_out": 0.183,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.042,
            "usd_per_m_out": 0.184,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.065,
            "usd_per_m_out": 0.24,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.24,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Tenstorrent",
            "tag": "tenstorrent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.25,
            "context_length": 262000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.084,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.0885,
            "usd_per_m_out": 0.295,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "OpenInference",
            "tag": "open-inference/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.108,
            "usd_per_m_out": 0.35,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.108,
            "usd_per_m_out": 0.35,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.1125,
            "usd_per_m_out": 0.375,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.38,
            "context_length": 1048576,
            "uptime_1d": 90.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.4,
            "context_length": 1048576,
            "uptime_1d": 98.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "NextBit",
            "tag": "nextbit/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.45,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.47,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Particle AI",
            "tag": "particle-ai/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "KT",
            "tag": "kt",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 96.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.625,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.625,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp4",
            "usd_per_m_in": 0.1875,
            "usd_per_m_out": 0.625,
            "context_length": 1048576,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 0.75,
            "context_length": 1048576,
            "uptime_1d": 95,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 42
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.3-Flash"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.08 per million output tokens."
    },
    {
      "model": "Step-3.7-173B-Reap-4.6bit-MLX",
      "made_by": "True2456",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/644f7e6233ac8f46fa0b9e26/CmF2ocXhkr2UtHXgmwq7-.png",
      "based_on": "Step 3.7 Flash",
      "variant_of": "/inferences/step-3-7/flash",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/step-3-7/flash~true2456-reap-4",
      "hub_url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash",
      "download_gb": 93,
      "params_b": 173,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "stepfun/step-3.7-flash",
        "url": "https://openrouter.ai/stepfun/step-3.7-flash",
        "input_usd_per_million": 0.2,
        "output_usd_per_million": 1.15,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 1.15,
            "context_length": 262144,
            "uptime_1d": 99.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StepFun",
            "tag": "stepfun/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 1.15,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/stepfun-ai/Step-3.7-Flash"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.15 per million output tokens."
    },
    {
      "model": "MiniMax-M2.7-RAM-100GB-MLX",
      "made_by": "baa-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/644d2651d185572dd1e497d9/-F6ZUGFzgXLlFa1asCLBW.png",
      "based_on": "MiniMax-M2.7 Base",
      "variant_of": "/inferences/minimax-m2-7/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/minimax-m2-7/base~baa-ai-ram-100gb",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7",
      "download_gb": 93.2,
      "params_b": 165,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m2.7",
        "url": "https://openrouter.ai/minimax/minimax-m2.7",
        "input_usd_per_million": 0.21,
        "output_usd_per_million": 0.84,
        "free_tier": false,
        "context_tokens": 196608,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 196608,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Groq",
            "tag": "groq",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.8,
            "context_length": 196608,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.84 per million output tokens."
    },
    {
      "model": "Step-3.7-173B-Ream-4.6bit-MLX",
      "made_by": "True2456",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/644f7e6233ac8f46fa0b9e26/CmF2ocXhkr2UtHXgmwq7-.png",
      "based_on": "Step 3.7 Flash",
      "variant_of": "/inferences/step-3-7/flash",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/step-3-7/flash~true2456-ream-4",
      "hub_url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash",
      "download_gb": 94.2,
      "params_b": 173,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "stepfun/step-3.7-flash",
        "url": "https://openrouter.ai/stepfun/step-3.7-flash",
        "input_usd_per_million": 0.2,
        "output_usd_per_million": 1.15,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 1.15,
            "context_length": 262144,
            "uptime_1d": 99.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StepFun",
            "tag": "stepfun/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 1.15,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/stepfun-ai/Step-3.7-Flash"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.15 per million output tokens."
    },
    {
      "model": "GLM-5.3-Flash-Coder-RCO-152E-GGUF",
      "made_by": "lapadatbogdan",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": "GLM-5.3 Flash",
      "variant_of": "/inferences/glm-5-3/flash",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-5-3/flash~lapadatbogdan-coder-rco-152e",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.3-Flash",
      "download_gb": 95.6,
      "params_b": 169,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.3-flash",
        "url": "https://openrouter.ai/z-ai/glm-5.3-flash",
        "input_usd_per_million": 0.0399,
        "output_usd_per_million": 0.0784,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.0399,
            "usd_per_m_out": 0.0784,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.065,
            "usd_per_m_out": 0.18,
            "context_length": 262144,
            "uptime_1d": 96.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.174,
            "usd_per_m_out": 0.183,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.042,
            "usd_per_m_out": 0.184,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.065,
            "usd_per_m_out": 0.24,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.24,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Tenstorrent",
            "tag": "tenstorrent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.25,
            "context_length": 262000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.084,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.0885,
            "usd_per_m_out": 0.295,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "OpenInference",
            "tag": "open-inference/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.108,
            "usd_per_m_out": 0.35,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.108,
            "usd_per_m_out": 0.35,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.1125,
            "usd_per_m_out": 0.375,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.38,
            "context_length": 1048576,
            "uptime_1d": 90.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.4,
            "context_length": 1048576,
            "uptime_1d": 98.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "NextBit",
            "tag": "nextbit/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.45,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.47,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Particle AI",
            "tag": "particle-ai/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "KT",
            "tag": "kt",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 96.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.625,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.625,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp4",
            "usd_per_m_in": 0.1875,
            "usd_per_m_out": 0.625,
            "context_length": 1048576,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 0.75,
            "context_length": 1048576,
            "uptime_1d": 95,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 42
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.3-Flash"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.08 per million output tokens."
    },
    {
      "model": "MiniMax-M2.7-RAM-120GB-MLX",
      "made_by": "baa-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/644d2651d185572dd1e497d9/-F6ZUGFzgXLlFa1asCLBW.png",
      "based_on": "MiniMax-M2.7 Base",
      "variant_of": "/inferences/minimax-m2-7/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/minimax-m2-7/base~baa-ai-ram-120gb",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7",
      "download_gb": 111.9,
      "params_b": 198,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m2.7",
        "url": "https://openrouter.ai/minimax/minimax-m2.7",
        "input_usd_per_million": 0.21,
        "output_usd_per_million": 0.84,
        "free_tier": false,
        "context_tokens": 196608,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 196608,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Groq",
            "tag": "groq",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.8,
            "context_length": 196608,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.84 per million output tokens."
    },
    {
      "model": "4xH100-Mistral-Medium-3.5-128B-MonEspaceSante",
      "made_by": "fenyo",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/634c17653d11eaedd88b314d/9OgyfKstSZtbmsmuG8MbU.png",
      "based_on": "Mistral Medium 3.5 128B",
      "variant_of": "/inferences/mistral-medium-3-5/128b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/mistral-medium-3-5/128b~fenyo-4xh100-monespacesante",
      "hub_url": "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B",
      "download_gb": 123.1,
      "params_b": 128,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "mistralai/mistral-medium-3-5",
        "url": "https://openrouter.ai/mistralai/mistral-medium-3-5",
        "input_usd_per_million": 1.5,
        "output_usd_per_million": 7.5,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": null,
            "usd_per_m_in": 1.5,
            "usd_per_m_out": 7.5,
            "context_length": 262144,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 1.65,
            "usd_per_m_out": 8.25,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/mistralai/Mistral-Medium-3.5-128B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $7.50 per million output tokens."
    },
    {
      "model": "MiniMax-M2.7-Abliterated-Heretic-GGUF",
      "made_by": "Youssofal",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6957877a8cedfe83dba58b8d/N_00rxRoECw72WQ8IObdd.png",
      "based_on": "MiniMax-M2.7 Base",
      "variant_of": "/inferences/minimax-m2-7/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/minimax-m2-7/base~youssofal-abliterated-heretic",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7",
      "download_gb": 128.8,
      "params_b": 228,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m2.7",
        "url": "https://openrouter.ai/minimax/minimax-m2.7",
        "input_usd_per_million": 0.21,
        "output_usd_per_million": 0.84,
        "free_tier": false,
        "context_tokens": 196608,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 196608,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Groq",
            "tag": "groq",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 1.8,
            "context_length": 196608,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 2
      },
      "licence": {
        "name": "custom licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/MiniMaxAI/MiniMax-M2.7"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.84 per million output tokens."
    },
    {
      "model": "DeepSeek-V4-Flash-0731-abliterated",
      "made_by": "prem-research",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": "DeepSeek V4 Flash-0731",
      "variant_of": "/inferences/deepseek-v4/flash-0731",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/deepseek-v4/flash-0731~prem-research-abliterated",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731",
      "download_gb": 155.4,
      "params_b": 157,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v4-flash-0731",
        "url": "https://openrouter.ai/deepseek/deepseek-v4-flash-0731",
        "input_usd_per_million": 0.0434,
        "output_usd_per_million": 0.3,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.14,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.048,
            "usd_per_m_out": 0.144,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.18,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/fast",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.23,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.119,
            "usd_per_m_out": 0.238,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.26,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp32",
            "quantization": "fp32",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.26,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.26,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.088,
            "usd_per_m_out": 0.264,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.09,
            "usd_per_m_out": 0.27,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.27,
            "context_length": 262144,
            "uptime_1d": 92.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 98.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.0434,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.42,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.495,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.176,
            "usd_per_m_out": 0.528,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/us-east-1",
            "quantization": null,
            "usd_per_m_in": 0.187,
            "usd_per_m_out": 0.561,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.22,
            "usd_per_m_out": 0.66,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.286,
            "usd_per_m_out": 0.858,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.4092,
            "usd_per_m_out": 1.2276,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.44,
            "usd_per_m_out": 1.32,
            "context_length": 1048576,
            "uptime_1d": null,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.44,
            "usd_per_m_out": 1.32,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.44,
            "usd_per_m_out": 1.32,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.44,
            "usd_per_m_out": 1.32,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.44,
            "usd_per_m_out": 1.32,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 29
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4-Flash-0731"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.30 per million output tokens."
    },
    {
      "model": "MiniMax-M3-ConfigI-MLX",
      "made_by": "thetom-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/676e38ad04af5bec20bc9faf/dUd-LsZEX0H_d4qefO_g6.jpeg",
      "based_on": "MiniMax-M3 Base",
      "variant_of": "/inferences/minimax-m3/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/minimax-m3/base~thetom-ai-configi",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M3",
      "download_gb": 155.6,
      "params_b": 276,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m3",
        "url": "https://openrouter.ai/minimax/minimax-m3",
        "input_usd_per_million": 0.3,
        "output_usd_per_million": 1.2,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.224,
            "usd_per_m_out": 0.88,
            "context_length": 524288,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.23,
            "usd_per_m_out": 0.96,
            "context_length": 262144,
            "uptime_1d": 91.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 0.96,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 1.1,
            "context_length": 524288,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 524288,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Minimax",
            "tag": "minimax",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 524288,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 8
      },
      "licence": {
        "name": "Minimax Community licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/MiniMaxAI/MiniMax-M3"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.20 per million output tokens."
    },
    {
      "model": "Hy3-Alis-MLX-Dynamic",
      "made_by": "avlp12",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5dd96eb166059660ed1ee413/Lp3m-XLpjQGwBItlvn69q.png",
      "based_on": "Hunyuan Hy3 Base",
      "variant_of": "/inferences/hy-3/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/hy-3/base~avlp12-alis-dynamic",
      "hub_url": "https://huggingface.co/tencent/Hy3",
      "download_gb": 156.9,
      "params_b": 278,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "tencent/hy3",
        "url": "https://openrouter.ai/tencent/hy3",
        "input_usd_per_million": 0.0825,
        "output_usd_per_million": 0.33,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Tencent",
            "tag": "tencent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.0825,
            "usd_per_m_out": 0.33,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.58,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.58,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2,
            "usd_per_m_out": 0.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 4
      },
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/tencent/Hy3"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.33 per million output tokens."
    },
    {
      "model": "MiMo-V2.5-Text-Only-oQ4-MLX",
      "made_by": "Blightbow",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/680cb7d1233834890a64acee/5w_4aLfF-7MAyaIPOV498.jpeg",
      "based_on": "MiMo V2.5 Base",
      "variant_of": "/inferences/mimo-v2-5/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/mimo-v2-5/base~blightbow-text-only-oq4",
      "hub_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.5",
      "download_gb": 164.6,
      "params_b": 292,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "xiaomi/mimo-v2.5",
        "url": "https://openrouter.ai/xiaomi/mimo-v2.5",
        "input_usd_per_million": 0.4,
        "output_usd_per_million": 2,
        "free_tier": false,
        "context_tokens": 1000000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.4,
            "usd_per_m_out": 2,
            "context_length": 1000000,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.5"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $2.00 per million output tokens."
    },
    {
      "model": "MiMo-V2.6-Flash-RL-Uncensored",
      "made_by": "dealignai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/699af30d096f7e8fe8d82a11/PEWX90_WdOgBjRvqhsHix.png",
      "based_on": "MiMo V2.6 Flash-RL",
      "variant_of": "/inferences/mimo-v2-6/flash-rl",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/mimo-v2-6/flash-rl~dealignai-uncensored",
      "hub_url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "download_gb": 165.5,
      "params_b": 167,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "xiaomi/mimo-v2.6-flash",
        "url": "https://openrouter.ai/xiaomi/mimo-v2.6-flash",
        "input_usd_per_million": 0.14,
        "output_usd_per_million": 0.28,
        "free_tier": false,
        "context_tokens": 1050000,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "GMICloud",
            "tag": "gmicloud/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.112,
            "usd_per_m_out": 0.224,
            "context_length": 1050000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 98.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Xiaomi",
            "tag": "xiaomi/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Tenstorrent",
            "tag": "tenstorrent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.165,
            "usd_per_m_out": 0.33,
            "context_length": 1048576,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.35,
            "context_length": 1000000,
            "uptime_1d": 97,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.36,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 8
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.28 per million output tokens."
    },
    {
      "model": "GLM-5.3-Flash-MLX-mixed-4_8bit",
      "made_by": "pipenetwork",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": "GLM-5.3 Flash",
      "variant_of": "/inferences/glm-5-3/flash",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-5-3/flash~pipenetwork-mixed-4",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.3-Flash",
      "download_gb": 169.4,
      "params_b": 300,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.3-flash",
        "url": "https://openrouter.ai/z-ai/glm-5.3-flash",
        "input_usd_per_million": 0.0399,
        "output_usd_per_million": 0.0784,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.0399,
            "usd_per_m_out": 0.0784,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Darkbloom",
            "tag": "darkbloom/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.065,
            "usd_per_m_out": 0.18,
            "context_length": 262144,
            "uptime_1d": 96.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.174,
            "usd_per_m_out": 0.183,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.042,
            "usd_per_m_out": 0.184,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.065,
            "usd_per_m_out": 0.24,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.24,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Tenstorrent",
            "tag": "tenstorrent/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.25,
            "context_length": 262000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.084,
            "usd_per_m_out": 0.28,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.0885,
            "usd_per_m_out": 0.295,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "OpenInference",
            "tag": "open-inference/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.108,
            "usd_per_m_out": 0.35,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.108,
            "usd_per_m_out": 0.35,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.1125,
            "usd_per_m_out": 0.375,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.38,
            "context_length": 1048576,
            "uptime_1d": 90.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.4,
            "context_length": 1048576,
            "uptime_1d": 98.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "NextBit",
            "tag": "nextbit/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.45,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.47,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Particle AI",
            "tag": "particle-ai/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "KT",
            "tag": "kt",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 96.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.625,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.175,
            "usd_per_m_out": 0.625,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp4",
            "usd_per_m_in": 0.1875,
            "usd_per_m_out": 0.625,
            "context_length": 1048576,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 0.75,
            "context_length": 1048576,
            "uptime_1d": 95,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 42
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.3-Flash"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.08 per million output tokens."
    },
    {
      "model": "GLM-5.2-Demolition-q4a4-soul-MLX",
      "made_by": "philipjohnbasile",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": "GLM-5.2 Base",
      "variant_of": "/inferences/glm-5-2/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-5-2/base~philipjohnbasile-demolition-soul",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.2",
      "download_gb": 174.6,
      "params_b": 309,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.2",
        "url": "https://openrouter.ai/z-ai/glm-5.2",
        "input_usd_per_million": 0.007,
        "output_usd_per_million": 6.8,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 0.88,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/flex",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.44,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.5625,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.643,
            "usd_per_m_out": 1.978,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ambient",
            "tag": "ambient/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2,
            "context_length": 202752,
            "uptime_1d": 50.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6496,
            "usd_per_m_out": 2.0416,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.651,
            "usd_per_m_out": 2.046,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 2.2,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.62,
            "usd_per_m_out": 2.42,
            "context_length": 1048576,
            "uptime_1d": 92.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/priority",
            "quantization": "fp4",
            "usd_per_m_in": 0.84375,
            "usd_per_m_out": 2.7,
            "context_length": 1048576,
            "uptime_1d": 98.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.938,
            "usd_per_m_out": 2.948,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.966,
            "usd_per_m_out": 3.036,
            "context_length": 1000000,
            "uptime_1d": 97.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8/us-east-1",
            "quantization": "fp8",
            "usd_per_m_in": 1.12,
            "usd_per_m_out": 3.52,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": "nvfp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8/us",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 262144,
            "uptime_1d": 98.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 1.54,
            "usd_per_m_out": 4.84,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/us/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.007,
            "usd_per_m_out": 6.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.31,
            "usd_per_m_out": 7.26,
            "context_length": 1000000,
            "uptime_1d": 92.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fast",
            "quantization": "fp4",
            "usd_per_m_in": 2.25,
            "usd_per_m_out": 7.88,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.0066,
            "usd_per_m_out": 7.9833,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer",
            "quantization": "fp4",
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 34
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.2"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $6.80 per million output tokens."
    },
    {
      "model": "DeepSeek-V4.1-Flash-1017-GSQ-agentic-strata",
      "made_by": "kaushikvira",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": "DeepSeek V4.1 Flash-1017",
      "variant_of": "/inferences/deepseek-v4-1/flash-1017",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/deepseek-v4-1/flash-1017~kaushikvira-gsq-agentic-strata",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "download_gb": 188.8,
      "params_b": 191,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v4.1-flash-1017",
        "url": "https://openrouter.ai/deepseek/deepseek-v4.1-flash-1017",
        "input_usd_per_million": 0.02,
        "output_usd_per_million": 0.108,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.02,
            "usd_per_m_out": 0.108,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.07272,
            "usd_per_m_out": 0.108,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream/flex",
            "quantization": null,
            "usd_per_m_in": 0.0775,
            "usd_per_m_out": 0.2325,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.0784,
            "usd_per_m_out": 0.2352,
            "context_length": 1048576,
            "uptime_1d": 96.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Particle AI",
            "tag": "particle-ai/us/flex",
            "quantization": null,
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.098,
            "usd_per_m_out": 0.294,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 0.0864,
            "usd_per_m_out": 0.3456,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StepFun",
            "tag": "stepfun/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.105,
            "usd_per_m_out": 0.42,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.105,
            "usd_per_m_out": 0.42,
            "context_length": 1000000,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.147,
            "usd_per_m_out": 0.441,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.111,
            "usd_per_m_out": 0.444,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.036,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Particle AI",
            "tag": "particle-ai/us",
            "quantization": null,
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.066,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.057,
            "usd_per_m_out": 0.506,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 0.55,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepSeek",
            "tag": "deepseek",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.605,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.605,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream",
            "quantization": null,
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.62,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.165,
            "usd_per_m_out": 0.66,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.68,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect",
            "quantization": null,
            "usd_per_m_in": 0.04368,
            "usd_per_m_out": 0.7,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.72,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.72,
            "context_length": 1048575,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.195,
            "usd_per_m_out": 0.78,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "OpenInference",
            "tag": "open-inference/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/spot",
            "quantization": null,
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 1048576,
            "uptime_1d": 98.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimaLabs",
            "tag": "primalabs",
            "quantization": null,
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 0.96,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.25,
            "usd_per_m_out": 1,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.16,
            "usd_per_m_out": 1.1,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 1.13,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.29,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 87.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fast",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 48
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.11 per million output tokens."
    },
    {
      "model": "GLM-5.2-REAP50-MLX-4bit",
      "made_by": "pipenetwork",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": "GLM-5.2 Base",
      "variant_of": "/inferences/glm-5-2/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-5-2/base~pipenetwork-reap50",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.2",
      "download_gb": 199.6,
      "params_b": 354,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.2",
        "url": "https://openrouter.ai/z-ai/glm-5.2",
        "input_usd_per_million": 0.007,
        "output_usd_per_million": 6.8,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 0.88,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/flex",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.44,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.5625,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.643,
            "usd_per_m_out": 1.978,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ambient",
            "tag": "ambient/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2,
            "context_length": 202752,
            "uptime_1d": 50.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6496,
            "usd_per_m_out": 2.0416,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.651,
            "usd_per_m_out": 2.046,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 2.2,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.62,
            "usd_per_m_out": 2.42,
            "context_length": 1048576,
            "uptime_1d": 92.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/priority",
            "quantization": "fp4",
            "usd_per_m_in": 0.84375,
            "usd_per_m_out": 2.7,
            "context_length": 1048576,
            "uptime_1d": 98.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.938,
            "usd_per_m_out": 2.948,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.966,
            "usd_per_m_out": 3.036,
            "context_length": 1000000,
            "uptime_1d": 97.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8/us-east-1",
            "quantization": "fp8",
            "usd_per_m_in": 1.12,
            "usd_per_m_out": 3.52,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": "nvfp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8/us",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 262144,
            "uptime_1d": 98.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 1.54,
            "usd_per_m_out": 4.84,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/us/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.007,
            "usd_per_m_out": 6.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.31,
            "usd_per_m_out": 7.26,
            "context_length": 1000000,
            "uptime_1d": 92.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fast",
            "quantization": "fp4",
            "usd_per_m_in": 2.25,
            "usd_per_m_out": 7.88,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.0066,
            "usd_per_m_out": 7.9833,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer",
            "quantization": "fp4",
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 34
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.2"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $6.80 per million output tokens."
    },
    {
      "model": "Sakura-MiniMax-M3.1-Flash-Q3_K_M-GGUF",
      "made_by": "webmp3",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/676e38ad04af5bec20bc9faf/dUd-LsZEX0H_d4qefO_g6.jpeg",
      "based_on": "MiniMax-M3.1 Flash",
      "variant_of": "/inferences/minimax-m3-1/flash",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/minimax-m3-1/flash~webmp3-sakura-k-m",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash",
      "download_gb": 240.4,
      "params_b": 426,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m3.1-flash",
        "url": "https://openrouter.ai/minimax/minimax-m3.1-flash",
        "input_usd_per_million": 0.075,
        "output_usd_per_million": 0.3,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Inferact vLLM",
            "tag": "inferact-vllm/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Minimax",
            "tag": "minimax",
            "quantization": null,
            "usd_per_m_in": 0.075,
            "usd_per_m_out": 0.3,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.135,
            "usd_per_m_out": 0.54,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/nvfp4",
            "quantization": "nvfp4",
            "usd_per_m_in": 0.135,
            "usd_per_m_out": 0.54,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SambaNova",
            "tag": "sambanova",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp4/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1000000,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 13
      },
      "licence": {
        "name": "Minimax Community licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $0.30 per million output tokens."
    },
    {
      "model": "Muse-Spark-Open-816B-A42B-lecore-mlx",
      "made_by": "staccs",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/64c115d1f661ace7c4022432/EbzfQisFJBEcKWDWLyV5O.png",
      "based_on": "Muse Spark Open 816B-A42B",
      "variant_of": "/inferences/muse-spark/816b-a42b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/muse-spark/816b-a42b~staccs-open-lecore",
      "hub_url": "https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B",
      "download_gb": 242.4,
      "params_b": 816,
      "active_params_b": 42,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "meta/muse-spark-open",
        "url": "https://openrouter.ai/meta/muse-spark-open",
        "input_usd_per_million": 1.25,
        "output_usd_per_million": 4.25,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 1.25,
            "usd_per_m_out": 4.25,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 1.25,
            "usd_per_m_out": 4.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.25,
            "usd_per_m_out": 4.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 1.25,
            "usd_per_m_out": 4.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.25,
            "usd_per_m_out": 4.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 5
      },
      "licence": {
        "name": "Muse Model License licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/meta-models/Muse-Spark-Open-816B-A42B"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $4.25 per million output tokens."
    },
    {
      "model": "MiniMax-M3-Mixed-4.5bit-MLX",
      "made_by": "Spangler3000",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/676e38ad04af5bec20bc9faf/dUd-LsZEX0H_d4qefO_g6.jpeg",
      "based_on": "MiniMax-M3 Base",
      "variant_of": "/inferences/minimax-m3/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/minimax-m3/base~spangler3000-mixed-4",
      "hub_url": "https://huggingface.co/MiniMaxAI/MiniMax-M3",
      "download_gb": 251.7,
      "params_b": 446,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "minimax/minimax-m3",
        "url": "https://openrouter.ai/minimax/minimax-m3",
        "input_usd_per_million": 0.3,
        "output_usd_per_million": 1.2,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.224,
            "usd_per_m_out": 0.88,
            "context_length": 524288,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.23,
            "usd_per_m_out": 0.96,
            "context_length": 262144,
            "uptime_1d": 91.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 0.96,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 1.1,
            "context_length": 524288,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 524288,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Minimax",
            "tag": "minimax",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 524288,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 8
      },
      "licence": {
        "name": "Minimax Community licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/MiniMaxAI/MiniMax-M3"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet. OpenRouter sells it at $1.20 per million output tokens."
    },
    {
      "model": "GLM-5.1-RAM-270GB-MLX",
      "made_by": "baa-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/644d2651d185572dd1e497d9/-F6ZUGFzgXLlFa1asCLBW.png",
      "based_on": "GLM-5.1 Base",
      "variant_of": "/inferences/glm-5-1/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-5-1/base~baa-ai-ram-270gb",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.1",
      "download_gb": 269.3,
      "params_b": 477,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.1",
        "url": "https://openrouter.ai/z-ai/glm-5.1",
        "input_usd_per_million": 1.4,
        "output_usd_per_million": 4.4,
        "free_tier": false,
        "context_tokens": 204800,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.9,
            "usd_per_m_out": 2.9,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.966,
            "usd_per_m_out": 3.036,
            "context_length": 200000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.19,
            "usd_per_m_out": 3.74,
            "context_length": 204800,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.26,
            "usd_per_m_out": 3.96,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.33,
            "usd_per_m_out": 4.18,
            "context_length": 202745,
            "uptime_1d": 95.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.38,
            "usd_per_m_out": 4.4,
            "context_length": 204800,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4014,
            "usd_per_m_out": 4.4044,
            "context_length": 200000,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 11
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.1"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 269.3 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $4.40 per million output tokens."
    },
    {
      "model": "GLM-5.2-Alis-MLX-Dynamic-3.5bpw",
      "made_by": "avlp12",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": "GLM-5.2 Base",
      "variant_of": "/inferences/glm-5-2/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-5-2/base~avlp12-alis-dynamic-3-5bpw",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.2",
      "download_gb": 310.1,
      "params_b": 549,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.2",
        "url": "https://openrouter.ai/z-ai/glm-5.2",
        "input_usd_per_million": 0.007,
        "output_usd_per_million": 6.8,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 0.88,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/flex",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.44,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.5625,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.643,
            "usd_per_m_out": 1.978,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ambient",
            "tag": "ambient/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2,
            "context_length": 202752,
            "uptime_1d": 50.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6496,
            "usd_per_m_out": 2.0416,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.651,
            "usd_per_m_out": 2.046,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 2.2,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.62,
            "usd_per_m_out": 2.42,
            "context_length": 1048576,
            "uptime_1d": 92.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/priority",
            "quantization": "fp4",
            "usd_per_m_in": 0.84375,
            "usd_per_m_out": 2.7,
            "context_length": 1048576,
            "uptime_1d": 98.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.938,
            "usd_per_m_out": 2.948,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.966,
            "usd_per_m_out": 3.036,
            "context_length": 1000000,
            "uptime_1d": 97.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8/us-east-1",
            "quantization": "fp8",
            "usd_per_m_in": 1.12,
            "usd_per_m_out": 3.52,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": "nvfp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8/us",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 262144,
            "uptime_1d": 98.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 1.54,
            "usd_per_m_out": 4.84,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/us/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.007,
            "usd_per_m_out": 6.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.31,
            "usd_per_m_out": 7.26,
            "context_length": 1000000,
            "uptime_1d": 92.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fast",
            "quantization": "fp4",
            "usd_per_m_in": 2.25,
            "usd_per_m_out": 7.88,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.0066,
            "usd_per_m_out": 7.9833,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer",
            "quantization": "fp4",
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 34
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.2"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 310.1 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $6.80 per million output tokens."
    },
    {
      "model": "Kimi-K2.5-MLX-2.8bit",
      "made_by": "spicyneuron",
      "maker_logo": "https://huggingface.co/avatars/1b90869be21e6376a955383e7427c5ac.svg",
      "based_on": "Kimi K2.5 Base",
      "variant_of": "/inferences/kimi-k2-5/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/kimi-k2-5/base~spicyneuron-2",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K2.5",
      "download_gb": 343,
      "params_b": 608,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k2.5",
        "url": "https://openrouter.ai/moonshotai/kimi-k2.5",
        "input_usd_per_million": 0.5,
        "output_usd_per_million": 2.5,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.49,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.5,
            "usd_per_m_out": 2.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.57,
            "usd_per_m_out": 2.85,
            "context_length": 262144,
            "uptime_1d": 99.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock/us-east-2",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 3,
            "context_length": 262144,
            "uptime_1d": 98.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice",
            "quantization": null,
            "usd_per_m_in": 0.532,
            "usd_per_m_out": 3.325,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 5
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K2.5"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 343 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $2.50 per million output tokens."
    },
    {
      "model": "DeepSeek-V3.2-RAM-350GB-MLX",
      "made_by": "baa-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/644d2651d185572dd1e497d9/-F6ZUGFzgXLlFa1asCLBW.png",
      "based_on": "DeepSeek V3.2 Base",
      "variant_of": "/inferences/deepseek-v3-2/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/deepseek-v3-2/base~baa-ai-ram-350gb",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V3.2",
      "download_gb": 350.5,
      "params_b": 621,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v3.2",
        "url": "https://openrouter.ai/deepseek/deepseek-v3.2",
        "input_usd_per_million": 0.26,
        "output_usd_per_million": 0.38,
        "free_tier": false,
        "context_tokens": 163840,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2088,
            "usd_per_m_out": 0.3096,
            "context_length": 163840,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.26,
            "usd_per_m_out": 0.38,
            "context_length": 163840,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.26,
            "usd_per_m_out": 0.38,
            "context_length": 163840,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 0.42,
            "context_length": 131072,
            "uptime_1d": 99.3,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.259,
            "usd_per_m_out": 0.8,
            "context_length": 163840,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 0.96,
            "context_length": 163840,
            "uptime_1d": 98.5,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 6
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V3.2"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 350.5 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $0.38 per million output tokens."
    },
    {
      "model": "GLM-5.1-RAM-420GB-MLX",
      "made_by": "baa-ai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/644d2651d185572dd1e497d9/-F6ZUGFzgXLlFa1asCLBW.png",
      "based_on": "GLM-5.1 Base",
      "variant_of": "/inferences/glm-5-1/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-5-1/base~baa-ai-ram-420gb",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.1",
      "download_gb": 355.6,
      "params_b": 630,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.1",
        "url": "https://openrouter.ai/z-ai/glm-5.1",
        "input_usd_per_million": 1.4,
        "output_usd_per_million": 4.4,
        "free_tier": false,
        "context_tokens": 204800,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.9,
            "usd_per_m_out": 2.9,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.966,
            "usd_per_m_out": 3.036,
            "context_length": 200000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.19,
            "usd_per_m_out": 3.74,
            "context_length": 204800,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.26,
            "usd_per_m_out": 3.96,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.33,
            "usd_per_m_out": 4.18,
            "context_length": 202745,
            "uptime_1d": 95.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.38,
            "usd_per_m_out": 4.4,
            "context_length": 204800,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4014,
            "usd_per_m_out": 4.4044,
            "context_length": 200000,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 11
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.1"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 355.6 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $4.40 per million output tokens."
    },
    {
      "model": "DeepSeek-V3.1-Terminus-MLX-mixed_4_6",
      "made_by": "finding1",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": "DeepSeek V3.1 Terminus",
      "variant_of": "/inferences/deepseek-v3-1/terminus",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/deepseek-v3-1/terminus~finding1-mixed-4-6",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V3.1-Terminus",
      "download_gb": 375.7,
      "params_b": 665,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v3.1-terminus",
        "url": "https://openrouter.ai/deepseek/deepseek-v3.1-terminus",
        "input_usd_per_million": 0.27,
        "output_usd_per_million": 1,
        "free_tier": false,
        "context_tokens": 163840,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.27,
            "usd_per_m_out": 1,
            "context_length": 163840,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V3.1-Terminus"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 375.7 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $1.00 per million output tokens."
    },
    {
      "model": "DeepSeek-V3.1-Terminus-MLX-mixed_4_6-v0.28.2",
      "made_by": "finding1",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": "DeepSeek V3.1 Terminus",
      "variant_of": "/inferences/deepseek-v3-1/terminus",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/deepseek-v3-1/terminus~finding1-mixed-4-6-28-2",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V3.1-Terminus",
      "download_gb": 375.8,
      "params_b": 666,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v3.1-terminus",
        "url": "https://openrouter.ai/deepseek/deepseek-v3.1-terminus",
        "input_usd_per_million": 0.27,
        "output_usd_per_million": 1,
        "free_tier": false,
        "context_tokens": 163840,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.27,
            "usd_per_m_out": 1,
            "context_length": 163840,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 1
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V3.1-Terminus"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 375.8 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $1.00 per million output tokens."
    },
    {
      "model": "GLM-5.1-MLX-4.5bit",
      "made_by": "spicyneuron",
      "maker_logo": "https://huggingface.co/avatars/1b90869be21e6376a955383e7427c5ac.svg",
      "based_on": "GLM-5.1 Base",
      "variant_of": "/inferences/glm-5-1/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-5-1/base~spicyneuron-4",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.1",
      "download_gb": 393,
      "params_b": 696,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.1",
        "url": "https://openrouter.ai/z-ai/glm-5.1",
        "input_usd_per_million": 1.4,
        "output_usd_per_million": 4.4,
        "free_tier": false,
        "context_tokens": 204800,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.9,
            "usd_per_m_out": 2.9,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.966,
            "usd_per_m_out": 3.036,
            "context_length": 200000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.19,
            "usd_per_m_out": 3.74,
            "context_length": 204800,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.26,
            "usd_per_m_out": 3.96,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.33,
            "usd_per_m_out": 4.18,
            "context_length": 202745,
            "uptime_1d": 95.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.38,
            "usd_per_m_out": 4.4,
            "context_length": 204800,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4014,
            "usd_per_m_out": 4.4044,
            "context_length": 200000,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 11
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.1"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 393 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $4.40 per million output tokens."
    },
    {
      "model": "Kimi-K2.6-MLX-3.3bit",
      "made_by": "spicyneuron",
      "maker_logo": "https://huggingface.co/avatars/1b90869be21e6376a955383e7427c5ac.svg",
      "based_on": "Kimi K2.6 Base",
      "variant_of": "/inferences/kimi-k2-6/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/kimi-k2-6/base~spicyneuron-3",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K2.6",
      "download_gb": 398,
      "params_b": 705,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k2.6",
        "url": "https://openrouter.ai/moonshotai/kimi-k2.6",
        "input_usd_per_million": 0.65,
        "output_usd_per_million": 3.41,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.39,
            "usd_per_m_out": 1.8,
            "context_length": 262144,
            "uptime_1d": 98.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.57,
            "usd_per_m_out": 2.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.5985,
            "usd_per_m_out": 2.52,
            "context_length": 256000,
            "uptime_1d": 98.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/flex",
            "quantization": "fp4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.77,
            "usd_per_m_out": 3.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.8,
            "usd_per_m_out": 3.4,
            "context_length": 262144,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.65,
            "usd_per_m_out": 3.41,
            "context_length": 262144,
            "uptime_1d": 94.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 91.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 256000,
            "uptime_1d": 91.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.7985,
            "usd_per_m_out": 3.75,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 73.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/priority",
            "quantization": "fp4",
            "usd_per_m_in": 1.125,
            "usd_per_m_out": 5.25,
            "context_length": 262144,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 17
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K2.6"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 398 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $3.41 per million output tokens."
    },
    {
      "model": "Kimi-K2.6-Reap-Solidity",
      "made_by": "sombra-x",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": "Kimi K2.6 Base",
      "variant_of": "/inferences/kimi-k2-6/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/kimi-k2-6/base~sombra-x-reap-solidity",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K2.6",
      "download_gb": 400.7,
      "params_b": 710,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k2.6",
        "url": "https://openrouter.ai/moonshotai/kimi-k2.6",
        "input_usd_per_million": 0.65,
        "output_usd_per_million": 3.41,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.39,
            "usd_per_m_out": 1.8,
            "context_length": 262144,
            "uptime_1d": 98.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.57,
            "usd_per_m_out": 2.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.5985,
            "usd_per_m_out": 2.52,
            "context_length": 256000,
            "uptime_1d": 98.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/flex",
            "quantization": "fp4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.77,
            "usd_per_m_out": 3.4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita",
            "quantization": null,
            "usd_per_m_in": 0.8,
            "usd_per_m_out": 3.4,
            "context_length": 262144,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.65,
            "usd_per_m_out": 3.41,
            "context_length": 262144,
            "uptime_1d": 94.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/bf16",
            "quantization": "bf16",
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 91.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 256000,
            "uptime_1d": 91.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.7985,
            "usd_per_m_out": 3.75,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 99.5,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 73.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/priority",
            "quantization": "fp4",
            "usd_per_m_in": 1.125,
            "usd_per_m_out": 5.25,
            "context_length": 262144,
            "uptime_1d": 99.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 17
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K2.6"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 400.7 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $3.41 per million output tokens."
    },
    {
      "model": "Kimi-K3-Audit-240E-MLX-MXFP4",
      "made_by": "catalystsec",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": "Kimi K3 BASE",
      "variant_of": "/inferences/kimi-k3/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/kimi-k3/base~catalystsec-audit-240e",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K3",
      "download_gb": 407.9,
      "params_b": 722,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k3",
        "url": "https://openrouter.ai/moonshotai/kimi-k3",
        "input_usd_per_million": 0.192,
        "output_usd_per_million": 16,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 1.85,
            "usd_per_m_out": 9.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.276,
            "usd_per_m_out": 9.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.1824,
            "usd_per_m_out": 11.0055,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 2.55,
            "usd_per_m_out": 12.95,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Io Net",
            "tag": "io-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 14,
            "context_length": 1048320,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 2.85,
            "usd_per_m_out": 14.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.234,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.234,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "KT",
            "tag": "kt",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock/us-east-2",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": null,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 98.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/fast",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.153,
            "usd_per_m_out": 16,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.192,
            "usd_per_m_out": 16,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.2,
            "usd_per_m_out": 16,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 3.45,
            "usd_per_m_out": 17.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us/fast",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 18,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fast",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 18,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp4",
            "usd_per_m_in": 4.5,
            "usd_per_m_out": 22.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 4.5,
            "usd_per_m_out": 22.5,
            "context_length": 1048576,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/fast",
            "quantization": null,
            "usd_per_m_in": 4.5,
            "usd_per_m_out": 22.5,
            "context_length": 1048576,
            "uptime_1d": 60.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 27
      },
      "licence": {
        "name": "Kimi K3 licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K3"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 407.9 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $16 per million output tokens."
    },
    {
      "model": "Kimi-K2.7-Code-MLX-3.6bit",
      "made_by": "spicyneuron",
      "maker_logo": "https://huggingface.co/avatars/1b90869be21e6376a955383e7427c5ac.svg",
      "based_on": "Kimi K2.7 Code",
      "variant_of": "/inferences/kimi-k2-7/code",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/kimi-k2-7/code~spicyneuron-3",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K2.7-Code",
      "download_gb": 427.6,
      "params_b": 757,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k2.7-code",
        "url": "https://openrouter.ai/moonshotai/kimi-k2.7-code",
        "input_usd_per_million": 0.7125,
        "output_usd_per_million": 3.5,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "CoreWeave",
            "tag": "coreweave/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 3,
            "context_length": 262144,
            "uptime_1d": 99.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake",
            "quantization": null,
            "usd_per_m_in": 0.7125,
            "usd_per_m_out": 3,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.85916,
            "usd_per_m_out": 3.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.912,
            "usd_per_m_out": 3.84,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai/highspeed",
            "quantization": null,
            "usd_per_m_in": 1.9,
            "usd_per_m_out": 8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 11
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K2.7-Code"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 427.6 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $3.50 per million output tokens."
    },
    {
      "model": "GLM-5.2-DQ4plus-q8",
      "made_by": "mlx-community",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/623c830997ddced06d78699b/3qTjC7d3YFCJTwpxd2noq.png",
      "based_on": "GLM-5.2 Base",
      "variant_of": "/inferences/glm-5-2/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-5-2/base~mlx-community-dq4plus",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.2",
      "download_gb": 432.7,
      "params_b": 766,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.2",
        "url": "https://openrouter.ai/z-ai/glm-5.2",
        "input_usd_per_million": 0.007,
        "output_usd_per_million": 6.8,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 0.88,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/flex",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.44,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.5625,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.643,
            "usd_per_m_out": 1.978,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ambient",
            "tag": "ambient/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2,
            "context_length": 202752,
            "uptime_1d": 50.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.6496,
            "usd_per_m_out": 2.0416,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.651,
            "usd_per_m_out": 2.046,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.7,
            "usd_per_m_out": 2.2,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.62,
            "usd_per_m_out": 2.42,
            "context_length": 1048576,
            "uptime_1d": 92.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp4/priority",
            "quantization": "fp4",
            "usd_per_m_in": 0.84375,
            "usd_per_m_out": 2.7,
            "context_length": 1048576,
            "uptime_1d": 98.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.938,
            "usd_per_m_out": 2.948,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.966,
            "usd_per_m_out": 3.036,
            "context_length": 1000000,
            "uptime_1d": 97.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8/us-east-1",
            "quantization": "fp8",
            "usd_per_m_in": 1.12,
            "usd_per_m_out": 3.52,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/zdr",
            "quantization": "nvfp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8/us",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048575,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 262144,
            "uptime_1d": 98.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/eu",
            "quantization": null,
            "usd_per_m_in": 1.54,
            "usd_per_m_out": 4.84,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/us/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 6.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.007,
            "usd_per_m_out": 6.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fast",
            "quantization": "fp8",
            "usd_per_m_in": 2.31,
            "usd_per_m_out": 7.26,
            "context_length": 1000000,
            "uptime_1d": 92.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fast",
            "quantization": "fp4",
            "usd_per_m_in": 2.25,
            "usd_per_m_out": 7.88,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.0066,
            "usd_per_m_out": 7.9833,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer",
            "quantization": "fp4",
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 34
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.2"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 432.7 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $6.80 per million output tokens."
    },
    {
      "model": "GLM-5.1-DQ4plus-q8",
      "made_by": "mlx-community",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/623c830997ddced06d78699b/3qTjC7d3YFCJTwpxd2noq.png",
      "based_on": "GLM-5.1 Base",
      "variant_of": "/inferences/glm-5-1/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-5-1/base~mlx-community-dq4plus",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.1",
      "download_gb": 433.2,
      "params_b": 767,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.1",
        "url": "https://openrouter.ai/z-ai/glm-5.1",
        "input_usd_per_million": 1.4,
        "output_usd_per_million": 4.4,
        "free_tier": false,
        "context_tokens": 204800,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.9,
            "usd_per_m_out": 2.9,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.966,
            "usd_per_m_out": 3.036,
            "context_length": 200000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.19,
            "usd_per_m_out": 3.74,
            "context_length": 204800,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.26,
            "usd_per_m_out": 3.96,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.33,
            "usd_per_m_out": 4.18,
            "context_length": 202745,
            "uptime_1d": 95.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.38,
            "usd_per_m_out": 4.4,
            "context_length": 204800,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4014,
            "usd_per_m_out": 4.4044,
            "context_length": 200000,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 11
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.1"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 433.2 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $4.40 per million output tokens."
    },
    {
      "model": "DeepSeek-V4.1-Flash-1017-abliterated",
      "made_by": "mrfakename",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": "DeepSeek V4.1 Flash-1017",
      "variant_of": "/inferences/deepseek-v4-1/flash-1017",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/deepseek-v4-1/flash-1017~mrfakename-abliterated",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "download_gb": 475.3,
      "params_b": 480,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v4.1-flash-1017",
        "url": "https://openrouter.ai/deepseek/deepseek-v4.1-flash-1017",
        "input_usd_per_million": 0.02,
        "output_usd_per_million": 0.108,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.02,
            "usd_per_m_out": 0.108,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.07272,
            "usd_per_m_out": 0.108,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream/flex",
            "quantization": null,
            "usd_per_m_in": 0.0775,
            "usd_per_m_out": 0.2325,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.0784,
            "usd_per_m_out": 0.2352,
            "context_length": 1048576,
            "uptime_1d": 96.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Particle AI",
            "tag": "particle-ai/us/flex",
            "quantization": null,
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.098,
            "usd_per_m_out": 0.294,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 0.0864,
            "usd_per_m_out": 0.3456,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StepFun",
            "tag": "stepfun/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.105,
            "usd_per_m_out": 0.42,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.105,
            "usd_per_m_out": 0.42,
            "context_length": 1000000,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.147,
            "usd_per_m_out": 0.441,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.111,
            "usd_per_m_out": 0.444,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.036,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Particle AI",
            "tag": "particle-ai/us",
            "quantization": null,
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.066,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.057,
            "usd_per_m_out": 0.506,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 0.55,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepSeek",
            "tag": "deepseek",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.605,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.605,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream",
            "quantization": null,
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.62,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.165,
            "usd_per_m_out": 0.66,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.68,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect",
            "quantization": null,
            "usd_per_m_in": 0.04368,
            "usd_per_m_out": 0.7,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.72,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.72,
            "context_length": 1048575,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.195,
            "usd_per_m_out": 0.78,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "OpenInference",
            "tag": "open-inference/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/spot",
            "quantization": null,
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 1048576,
            "uptime_1d": 98.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimaLabs",
            "tag": "primalabs",
            "quantization": null,
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 0.96,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.25,
            "usd_per_m_out": 1,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.16,
            "usd_per_m_out": 1.1,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 1.13,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.29,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 87.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fast",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 48
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 475.3 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $0.11 per million output tokens."
    },
    {
      "model": "DeepSeek-V4.1-Flash-1017-FP8-GGUF",
      "made_by": "AMAImedia",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/6538815d1bdb3c40db94fbfa/xMBly9PUMphrFVMxLX4kq.png",
      "based_on": "DeepSeek V4.1 Flash",
      "variant_of": "/inferences/deepseek-v4-1/flash",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/deepseek-v4-1/flash~amaimedia-1017",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash",
      "download_gb": 475.3,
      "params_b": 480,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v4.1-flash",
        "url": "https://openrouter.ai/deepseek/deepseek-v4.1-flash",
        "input_usd_per_million": 0.0507,
        "output_usd_per_million": 0.2028,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.112,
            "usd_per_m_out": 0.336,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.105,
            "usd_per_m_out": 0.42,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.42,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.48,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.135,
            "usd_per_m_out": 0.54,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.051,
            "usd_per_m_out": 0.567,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.147,
            "usd_per_m_out": 0.588,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp8",
            "usd_per_m_in": 0.103,
            "usd_per_m_out": 0.6,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.103,
            "usd_per_m_out": 0.6,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.63,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.64,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.65,
            "context_length": 1048576,
            "uptime_1d": 92.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 0.65,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.195,
            "usd_per_m_out": 0.66,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.72,
            "context_length": 1048575,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2125,
            "usd_per_m_out": 0.75,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.195,
            "usd_per_m_out": 0.78,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "OpenInference",
            "tag": "open-inference/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/spot",
            "quantization": null,
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 0.9,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.26,
            "usd_per_m_out": 1.17,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.29,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Google",
            "tag": "google-vertex/global",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/us",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/ultrafast",
            "quantization": "fp8",
            "usd_per_m_in": 3,
            "usd_per_m_out": 12,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 36
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 475.3 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $0.20 per million output tokens."
    },
    {
      "model": "DeepSeek-V4.1-Flash-1017-Uncensored-FP8",
      "made_by": "dealignai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/699af30d096f7e8fe8d82a11/PEWX90_WdOgBjRvqhsHix.png",
      "based_on": "DeepSeek V4.1 Flash-1017",
      "variant_of": "/inferences/deepseek-v4-1/flash-1017",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/deepseek-v4-1/flash-1017~dealignai-uncensored",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "download_gb": 475.3,
      "params_b": 480,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v4.1-flash-1017",
        "url": "https://openrouter.ai/deepseek/deepseek-v4.1-flash-1017",
        "input_usd_per_million": 0.02,
        "output_usd_per_million": 0.108,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.02,
            "usd_per_m_out": 0.108,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Sail Research",
            "tag": "sail-research/fp4/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.07272,
            "usd_per_m_out": 0.108,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream/flex",
            "quantization": null,
            "usd_per_m_in": 0.0775,
            "usd_per_m_out": 0.2325,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.0784,
            "usd_per_m_out": 0.2352,
            "context_length": 1048576,
            "uptime_1d": 96.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Particle AI",
            "tag": "particle-ai/us/flex",
            "quantization": null,
            "usd_per_m_in": 0.03,
            "usd_per_m_out": 0.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.098,
            "usd_per_m_out": 0.294,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 0.0864,
            "usd_per_m_out": 0.3456,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StepFun",
            "tag": "stepfun/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.105,
            "usd_per_m_out": 0.42,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.105,
            "usd_per_m_out": 0.42,
            "context_length": 1000000,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.147,
            "usd_per_m_out": 0.441,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.111,
            "usd_per_m_out": 0.444,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.036,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Particle AI",
            "tag": "particle-ai/us",
            "quantization": null,
            "usd_per_m_in": 0.06,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.066,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Morph",
            "tag": "morph/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.057,
            "usd_per_m_out": 0.506,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 0.55,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.1,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepSeek",
            "tag": "deepseek",
            "quantization": null,
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 0.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.605,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.08,
            "usd_per_m_out": 0.605,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Ionstream",
            "tag": "ionstream",
            "quantization": null,
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.62,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.165,
            "usd_per_m_out": 0.66,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DekaLLM",
            "tag": "dekallm",
            "quantization": null,
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.68,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimeIntellect",
            "tag": "primeintellect",
            "quantization": null,
            "usd_per_m_in": 0.04368,
            "usd_per_m_out": 0.7,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.72,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.72,
            "context_length": 1048575,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.195,
            "usd_per_m_out": 0.78,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "OpenInference",
            "tag": "open-inference/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/spot",
            "quantization": null,
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 1048576,
            "uptime_1d": 98.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "PrimaLabs",
            "tag": "primalabs",
            "quantization": null,
            "usd_per_m_in": 0.24,
            "usd_per_m_out": 0.96,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.25,
            "usd_per_m_out": 1,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.16,
            "usd_per_m_out": 1.1,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.28,
            "usd_per_m_out": 1.13,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.15,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.29,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mistral",
            "tag": "mistral/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 87.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Near AI",
            "tag": "near-ai",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fast",
            "quantization": null,
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 2.4,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 48
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 475.3 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $0.11 per million output tokens."
    },
    {
      "model": "DeepSeek-V4.1-Flash-Uncensored-FP8",
      "made_by": "dealignai",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/699af30d096f7e8fe8d82a11/PEWX90_WdOgBjRvqhsHix.png",
      "based_on": "DeepSeek V4.1 Flash",
      "variant_of": "/inferences/deepseek-v4-1/flash",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/deepseek-v4-1/flash~dealignai-uncensored",
      "hub_url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash",
      "download_gb": 475.3,
      "params_b": 480,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "deepseek/deepseek-v4.1-flash",
        "url": "https://openrouter.ai/deepseek/deepseek-v4.1-flash",
        "input_usd_per_million": 0.0507,
        "output_usd_per_million": 0.2028,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/flex",
            "quantization": "fp8",
            "usd_per_m_in": 0.112,
            "usd_per_m_out": 0.336,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 0.105,
            "usd_per_m_out": 0.42,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.14,
            "usd_per_m_out": 0.42,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.48,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.135,
            "usd_per_m_out": 0.54,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.051,
            "usd_per_m_out": 0.567,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.147,
            "usd_per_m_out": 0.588,
            "context_length": 1024000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp8",
            "usd_per_m_in": 0.103,
            "usd_per_m_out": 0.6,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.103,
            "usd_per_m_out": 0.6,
            "context_length": 1040000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/fp8/priority",
            "quantization": "fp8",
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.63,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Reka",
            "tag": "reka",
            "quantization": null,
            "usd_per_m_in": 0.12,
            "usd_per_m_out": 0.64,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "CoreWeave",
            "tag": "coreweave/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.13,
            "usd_per_m_out": 0.65,
            "context_length": 1048576,
            "uptime_1d": 92.6,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.19,
            "usd_per_m_out": 0.65,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cohere",
            "tag": "cohere/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.195,
            "usd_per_m_out": 0.66,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.18,
            "usd_per_m_out": 0.72,
            "context_length": 1048575,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Mancer 2",
            "tag": "mancer/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.2125,
            "usd_per_m_out": 0.75,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.195,
            "usd_per_m_out": 0.78,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "OpenInference",
            "tag": "open-inference/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.07,
            "usd_per_m_out": 0.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/spot",
            "quantization": null,
            "usd_per_m_in": 0.21,
            "usd_per_m_out": 0.84,
            "context_length": 1048576,
            "uptime_1d": 99.4,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 0.225,
            "usd_per_m_out": 0.9,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.26,
            "usd_per_m_out": 1.17,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Crusoe",
            "tag": "crusoe/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.29,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Google",
            "tag": "google-vertex/global",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/us",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1048576,
            "uptime_1d": 99.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.3,
            "usd_per_m_out": 1.2,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp8",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 1.8,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/ultrafast",
            "quantization": "fp8",
            "usd_per_m_in": 3,
            "usd_per_m_out": 12,
            "context_length": 1000000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 36
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 475.3 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $0.20 per million output tokens."
    },
    {
      "model": "Kimi-K2.7-Code-MLX-4bit-hiprec",
      "made_by": "pipenetwork",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": "Kimi K2.7 Code",
      "variant_of": "/inferences/kimi-k2-7/code",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/kimi-k2-7/code~pipenetwork-hiprec",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K2.7-Code",
      "download_gb": 599.7,
      "params_b": 1062,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k2.7-code",
        "url": "https://openrouter.ai/moonshotai/kimi-k2.7-code",
        "input_usd_per_million": 0.7125,
        "output_usd_per_million": 3.5,
        "free_tier": false,
        "context_tokens": 262144,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "CoreWeave",
            "tag": "coreweave/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 3,
            "context_length": 262144,
            "uptime_1d": 99.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake",
            "quantization": null,
            "usd_per_m_in": 0.7125,
            "usd_per_m_out": 3,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Inceptron",
            "tag": "inceptron/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.6,
            "usd_per_m_out": 3.5,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.75,
            "usd_per_m_out": 3.5,
            "context_length": 256000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.85916,
            "usd_per_m_out": 3.8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/int4",
            "quantization": "int4",
            "usd_per_m_in": 0.912,
            "usd_per_m_out": 3.84,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Cloudflare",
            "tag": "cloudflare",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai",
            "quantization": null,
            "usd_per_m_in": 0.95,
            "usd_per_m_out": 4,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai/highspeed",
            "quantization": null,
            "usd_per_m_in": 1.9,
            "usd_per_m_out": 8,
            "context_length": 262144,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 11
      },
      "licence": {
        "name": "Modified MIT licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K2.7-Code"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 599.7 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $3.50 per million output tokens."
    },
    {
      "model": "GLM-5.1-Abliterated",
      "made_by": "helixdouble",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/62dc173789b4cf157d36ebee/i_pxzM2ZDo3Ub-BEgIkE9.png",
      "based_on": "GLM-5.1 Base",
      "variant_of": "/inferences/glm-5-1/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/glm-5-1/base~helixdouble-abliterated",
      "hub_url": "https://huggingface.co/zai-org/GLM-5.1-FP8",
      "download_gb": 704.2,
      "params_b": 712,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "z-ai/glm-5.1",
        "url": "https://openrouter.ai/z-ai/glm-5.1",
        "input_usd_per_million": 1.4,
        "output_usd_per_million": 4.4,
        "free_tier": false,
        "context_tokens": 204800,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.9,
            "usd_per_m_out": 2.9,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "StreamLake",
            "tag": "streamlake/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 0.966,
            "usd_per_m_out": 3.036,
            "context_length": 200000,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "SiliconFlow",
            "tag": "siliconflow/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.19,
            "usd_per_m_out": 3.74,
            "context_length": 204800,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "AtlasCloud",
            "tag": "atlas-cloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.26,
            "usd_per_m_out": 3.96,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.33,
            "usd_per_m_out": 4.18,
            "context_length": 202745,
            "uptime_1d": 95.2,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Novita",
            "tag": "novita/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.38,
            "usd_per_m_out": 4.4,
            "context_length": 204800,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Baidu",
            "tag": "baidu/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "GMICloud",
            "tag": "gmicloud/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Friendli",
            "tag": "friendli",
            "quantization": null,
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Z.AI",
            "tag": "z-ai/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4,
            "usd_per_m_out": 4.4,
            "context_length": 202752,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Venice",
            "tag": "venice/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 1.4014,
            "usd_per_m_out": 4.4044,
            "context_length": 200000,
            "uptime_1d": 99,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 11
      },
      "licence": {
        "name": "MIT",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/zai-org/GLM-5.1-FP8"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 704.2 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $4.40 per million output tokens."
    },
    {
      "model": "Kimi-K3-Audit-451E-MLX-MXFP4",
      "made_by": "catalystsec",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": "Kimi K3 BASE",
      "variant_of": "/inferences/kimi-k3/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/kimi-k3/base~catalystsec-audit-451e",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K3",
      "download_gb": 725.3,
      "params_b": 1285,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k3",
        "url": "https://openrouter.ai/moonshotai/kimi-k3",
        "input_usd_per_million": 0.192,
        "output_usd_per_million": 16,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 1.85,
            "usd_per_m_out": 9.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.276,
            "usd_per_m_out": 9.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.1824,
            "usd_per_m_out": 11.0055,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 2.55,
            "usd_per_m_out": 12.95,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Io Net",
            "tag": "io-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 14,
            "context_length": 1048320,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 2.85,
            "usd_per_m_out": 14.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.234,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.234,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "KT",
            "tag": "kt",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock/us-east-2",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": null,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 98.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/fast",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.153,
            "usd_per_m_out": 16,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.192,
            "usd_per_m_out": 16,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.2,
            "usd_per_m_out": 16,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 3.45,
            "usd_per_m_out": 17.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us/fast",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 18,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fast",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 18,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp4",
            "usd_per_m_in": 4.5,
            "usd_per_m_out": 22.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 4.5,
            "usd_per_m_out": 22.5,
            "context_length": 1048576,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/fast",
            "quantization": null,
            "usd_per_m_in": 4.5,
            "usd_per_m_out": 22.5,
            "context_length": 1048576,
            "uptime_1d": 60.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 27
      },
      "licence": {
        "name": "Kimi K3 licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K3"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 725.3 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $16 per million output tokens."
    },
    {
      "model": "Kimi-K3-2bit-Uvmax",
      "made_by": "kernelpool",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/641c1e77c3983aa9490f8121/X1yT2rsaIbR9cdYGEVu0X.jpeg",
      "based_on": "Kimi K3 BASE",
      "variant_of": "/inferences/kimi-k3/base",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/kimi-k3/base~kernelpool-uvmax",
      "hub_url": "https://huggingface.co/moonshotai/Kimi-K3",
      "download_gb": 760.7,
      "params_b": 1347,
      "active_params_b": null,
      "params_basis": "estimated from download size",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": true,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [],
      "measured_on": [],
      "hosted_api": {
        "seller": "OpenRouter",
        "model_id": "moonshotai/kimi-k3",
        "url": "https://openrouter.ai/moonshotai/kimi-k3",
        "input_usd_per_million": 0.192,
        "output_usd_per_million": 16,
        "free_tier": false,
        "context_tokens": 1048576,
        "checked_at": "2026-11-15T19:26:02.086Z",
        "providers": [
          {
            "provider": "Chutes",
            "tag": "chutes/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 1.85,
            "usd_per_m_out": 9.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/us",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.276,
            "usd_per_m_out": 9.6,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Relace",
            "tag": "relace/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.1824,
            "usd_per_m_out": 11.0055,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DigitalOcean",
            "tag": "digitalocean",
            "quantization": null,
            "usd_per_m_in": 2.55,
            "usd_per_m_out": 12.95,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Io Net",
            "tag": "io-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 2.1,
            "usd_per_m_out": 14,
            "context_length": 1048320,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "DeepInfra",
            "tag": "deepinfra/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 2.85,
            "usd_per_m_out": 14.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us",
            "quantization": "fp4",
            "usd_per_m_in": 0.234,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 0.234,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Nebius",
            "tag": "nebius/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "River",
            "tag": "river",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "KT",
            "tag": "kt",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Amazon Bedrock",
            "tag": "amazon-bedrock/us-east-2",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Modal",
            "tag": "modal/mxfp4",
            "quantization": "mxfp4",
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": null,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Together",
            "tag": "together",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 98.1,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "BaseTen",
            "tag": "baseten/fp8",
            "quantization": "fp8",
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Moonshot AI",
            "tag": "moonshotai",
            "quantization": null,
            "usd_per_m_in": 3,
            "usd_per_m_out": 15,
            "context_length": 1048576,
            "uptime_1d": 99.9,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer/fast",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.153,
            "usd_per_m_out": 16,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Wafer",
            "tag": "wafer",
            "quantization": "mxfp4",
            "usd_per_m_in": 0.192,
            "usd_per_m_out": 16,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Makora",
            "tag": "makora/fp4",
            "quantization": "fp4",
            "usd_per_m_in": 1.2,
            "usd_per_m_out": 16,
            "context_length": 1048576,
            "uptime_1d": 99.8,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Alibaba",
            "tag": "alibaba",
            "quantization": null,
            "usd_per_m_in": 3.45,
            "usd_per_m_out": 17.25,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/us/fast",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 18,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "InferenceNet",
            "tag": "inference-net/fast",
            "quantization": "fp4",
            "usd_per_m_in": 0.45,
            "usd_per_m_out": 18,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Parasail",
            "tag": "parasail/fast",
            "quantization": "fp4",
            "usd_per_m_in": 4.5,
            "usd_per_m_out": 22.5,
            "context_length": 1048576,
            "uptime_1d": 100,
            "tools": false,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/us",
            "quantization": null,
            "usd_per_m_in": 4.5,
            "usd_per_m_out": 22.5,
            "context_length": 1048576,
            "uptime_1d": 98.7,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          },
          {
            "provider": "Fireworks",
            "tag": "fireworks/fast",
            "quantization": null,
            "usd_per_m_in": 4.5,
            "usd_per_m_out": 22.5,
            "context_length": 1048576,
            "uptime_1d": 60.3,
            "tools": true,
            "latency_ms": null,
            "tokens_per_sec": null
          }
        ],
        "provider_count": 27
      },
      "licence": {
        "name": "Kimi K3 licence",
        "you_may_serve": "Only your own machines and your company's",
        "class": "restricted",
        "source": "https://huggingface.co/moonshotai/Kimi-K3"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not run here yet: it needs about 760.7 GB of memory and the largest machine on the network has 256 GB. OpenRouter sells it at $16 per million output tokens."
    },
    {
      "model": "ernie-4.5-moe-tiny-random",
      "made_by": "yujiepan",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/64f187a2cc1c03340ac30498/TYYUxK8xD1AxExFMWqbZD.png",
      "based_on": "ERNIE 4.5 21B-A3B-THINKING",
      "variant_of": "/inferences/ernie-4-5/21b-a3b-thinking",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/ernie-4-5/21b-a3b-thinking~yujiepan-moe-tiny-random",
      "hub_url": "https://huggingface.co/baidu/ERNIE-4.5-21B-A3B-Thinking",
      "download_gb": 0,
      "params_b": 21,
      "active_params_b": 3,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [
        "English"
      ],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/baidu/ERNIE-4.5-21B-A3B-Thinking"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet."
    },
    {
      "model": "ernie-4.5-tiny-random",
      "made_by": "yujiepan",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/64f187a2cc1c03340ac30498/TYYUxK8xD1AxExFMWqbZD.png",
      "based_on": "ERNIE 4.5 0.3B",
      "variant_of": "/inferences/ernie-4-5/0.3b",
      "in_catalog": true,
      "kind": "Text & chat",
      "community_build": true,
      "page": "/inferences/ernie-4-5/0.3b~yujiepan-tiny-random",
      "hub_url": "https://huggingface.co/baidu/ERNIE-4.5-0.3B-PT",
      "download_gb": 0,
      "params_b": 0.3,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [],
      "languages_passed": [
        "English"
      ],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [
        "2026-W44"
      ],
      "measured_on": [],
      "hosted_api": null,
      "licence": {
        "name": "Apache 2.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/baidu/ERNIE-4.5-0.3B-PT"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet."
    },
    {
      "model": "wespeaker-voxceleb-resnet34-LM",
      "made_by": "pyannote",
      "maker_logo": "https://cdn-avatars.huggingface.co/v1/production/uploads/5fcf602140b27fad857dfa8d/-afjRExJpYCCl5TCtFA68.png",
      "based_on": null,
      "variant_of": null,
      "in_catalog": true,
      "kind": "Speaker recognition",
      "community_build": false,
      "page": "/inferences/wespeaker/resnet34-lm",
      "hub_url": "https://huggingface.co/pyannote/wespeaker-voxceleb-resnet34-LM",
      "download_gb": 0,
      "params_b": 0.01,
      "active_params_b": null,
      "params_basis": "named",
      "longest_prompt_tokens": null,
      "too_big_for_the_network": false,
      "largest_machine_gb": 256,
      "skills_passed": [
        "identifies speakers"
      ],
      "languages_passed": [],
      "status": "not measured yet",
      "fastest": null,
      "cheapest": null,
      "machine_kinds": 0,
      "machines_total": 0,
      "weeks_measured": [
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "measured_on": [],
      "hosted_api": null,
      "licence": {
        "name": "CC BY 4.0",
        "you_may_serve": "Anyone, including strangers on the open network",
        "class": "open",
        "source": "https://huggingface.co/pyannote/wespeaker-voxceleb-resnet34-LM"
      },
      "builds": 1,
      "variants": [],
      "bottom_line": "Not measured here yet."
    }
  ],
  "rows": [
    {
      "ref": "AEmotionStudio/stable-audio-open-models",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "AEmotionStudio/stable-audio-open-models",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W36"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W36"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-05T17:39:38.306Z",
          "last_at": "2026-09-05T17:39:38.306Z",
          "caps": [],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "AMAImedia/DeepSeek-V4.1-Flash-1017-FP8-GGUF",
      "family": "deepseek-v4-1",
      "member": "flash~amaimedia-1017",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4.1-Flash",
      "hub_repo": "deepseek-ai/DeepSeek-V4.1-Flash",
      "publisher": "AMAImedia",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 486656,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4.1-Flash",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4.1-flash",
        "ids": [
          "deepseek/deepseek-v4.1-flash",
          "deepseek/deepseek-v4.1-flash:batch"
        ],
        "usd_per_m_in": 0.0507,
        "usd_per_m_out": 0.2028,
        "usd_per_m_cached_in": 0.001014,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.336,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 488.0475,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "AMAImedia/MiniMax-M3.1-Flash-NVFP4-GGUF:Q4_K_M",
      "family": "minimax-m3-1",
      "member": "flash~lyjonathan",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "LyJonathan/MiniMax-M3.1-Flash-GGUF",
      "hub_repo": "LyJonathan/MiniMax-M3.1-Flash-GGUF",
      "publisher": "AMAImedia",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 246138,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "minimax-community",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/LyJonathan/MiniMax-M3.1-Flash-GGUF",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m3.1-flash",
        "ids": [
          "minimax/minimax-m3.1-flash"
        ],
        "usd_per_m_in": 0.075,
        "usd_per_m_out": 0.3,
        "usd_per_m_cached_in": 0.0075,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "AlphaOxO/GLM-OCR-NVFP4",
      "family": "glm-ocr",
      "member": "base",
      "quant": "W4A16",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-OCR",
      "hub_repo": "zai-org/GLM-OCR",
      "publisher": "AlphaOxO",
      "derivative": false,
      "cohort": 1,
      "rung": 1,
      "disk_mib": 1301,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-OCR",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 0.076,
          "first_at": "2026-11-03T21:08:21.853Z",
          "last_at": "2026-11-03T21:08:21.853Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output",
            "vision",
            "vision.read"
          ],
          "est_tokens_per_sec": 387.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2904,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "Ares-Realm-Studios/Mistral-Small-4-119B-2603",
      "family": "mistral-small-4",
      "member": "119b-2603~ares-realm-studios",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": null,
      "hub_repo": "Ares-Realm-Studios/Mistral-Small-4-119B-2603",
      "publisher": "Ares-Realm-Studios",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 230646,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Ares-Realm-Studios/Mistral-Small-4-119B-2603",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-small-2603",
        "ids": [
          "mistralai/mistral-small-2603"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": 0.015,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 136.6338,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "AutomatosX/AX-Hy-MT2-30B-A3B-MLX-AXQ-MXFP4",
      "family": "hy-mt-2",
      "member": "30b-a3b~automatosx-ax-axq",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "tencent/Hy-MT2-30B-A3B",
      "hub_repo": "tencent/Hy-MT2-30B-A3B",
      "publisher": "AutomatosX",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 15504,
      "active_share": 0.1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy-MT2-30B-A3B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy-mt2-30b-a3b",
        "ids": [
          "tencent/hy-mt2-30b-a3b"
        ],
        "usd_per_m_in": 0.074,
        "usd_per_m_out": 0.295,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "Beh1222/Sana_600M_1024px_diffusers",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "Beh1222/Sana_600M_1024px_diffusers",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Beh1222/Sana_600M_1024px_diffusers",
        "fetched_at": "2026-11-09T08:49:44.679Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:23:26.066Z",
          "last_at": "2026-11-03T22:23:26.066Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "Blightbow/MiMo-V2.5-Text-Only-oQ4-MLX",
      "family": "mimo-v2-5",
      "member": "base~blightbow-text-only-oq4",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "XiaomiMiMo/MiMo-V2.5",
      "hub_repo": "XiaomiMiMo/MiMo-V2.5",
      "publisher": "Blightbow",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 168539,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.5",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.5",
        "ids": [
          "xiaomi/mimo-v2.5"
        ],
        "usd_per_m_in": 0.4,
        "usd_per_m_out": 2,
        "usd_per_m_cached_in": 0.08,
        "context_length": 1000000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "CohereLabs/North-Mini-Code-1.0-fp8",
      "family": "north-1",
      "member": "mini-code-1-0",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "CohereLabs/North-Mini-Code-1.0",
      "hub_repo": "CohereLabs/North-Mini-Code-1.0",
      "publisher": "CohereLabs",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 30542,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/CohereLabs/North-Mini-Code-1.0",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.2606,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "Darylgeero/sd-turbo",
      "family": "sd-turbo",
      "member": "base",
      "quant": "FP16",
      "engine": "diffusers",
      "modality": "image",
      "base": null,
      "hub_repo": "Darylgeero/sd-turbo",
      "publisher": "Darylgeero",
      "derivative": false,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 12354,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Darylgeero/sd-turbo",
        "fetched_at": "2026-11-12T05:10:06.752Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-30T17:52:12.506Z",
          "last_at": "2026-10-30T17:52:12.506Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "DevQuasar/moonshotai.Kimi-K2.6-GGUF:Q4_K_M",
      "family": "kimi-k2-6",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.6",
      "hub_repo": "moonshotai/Kimi-K2.6",
      "publisher": "DevQuasar",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 591807,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.6",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.6",
        "ids": [
          "moonshotai/kimi-k2.6"
        ],
        "usd_per_m_in": 0.65,
        "usd_per_m_out": 3.41,
        "usd_per_m_cached_in": 0.15,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "DevQuasar/moonshotai.Kimi-K2.7-Code-GGUF:Q4_K_M",
      "family": "kimi-k2-7",
      "member": "code",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.7-Code",
      "hub_repo": "moonshotai/Kimi-K2.7-Code",
      "publisher": "DevQuasar",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 591810,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.7-Code",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.7-code",
        "ids": [
          "moonshotai/kimi-k2.7-code"
        ],
        "usd_per_m_in": 0.7125,
        "usd_per_m_out": 3.5,
        "usd_per_m_cached_in": 0.16,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "Efficient-Large-Model/Sana_1600M_1024px_diffusers",
      "family": "sana",
      "member": "1600m-1024px",
      "quant": "FP16",
      "engine": "diffusers",
      "modality": "image",
      "base": null,
      "hub_repo": "Efficient-Large-Model/Sana_1600M_1024px_diffusers",
      "publisher": "Efficient-Large-Model",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 3060,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Efficient-Large-Model/Sana_1600M_1024px_diffusers",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W36",
        "2026-W37",
        "2026-W38",
        "2026-W43",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T07:05:09.990Z",
          "last_at": "2026-10-22T08:18:05.273Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:22:32.932Z",
          "last_at": "2026-11-03T22:22:32.932Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-13T18:37:04.807Z",
          "last_at": "2026-09-13T18:37:04.807Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W36"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-02T19:21:32.909Z",
          "last_at": "2026-09-02T19:21:32.909Z",
          "caps": [],
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-14T04:29:46.125Z",
          "last_at": "2026-09-14T04:29:46.125Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "EmbeddedLLM/MiniMax-M3-FP8-dynamic",
      "family": "minimax-m3",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M3",
      "hub_repo": "MiniMaxAI/MiniMax-M3",
      "publisher": "EmbeddedLLM",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 410739,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "minimax-community",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M3",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m3",
        "ids": [
          "minimax/minimax-m3",
          "minimax/minimax-m3:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.06,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.88,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 362.4293,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "Harahan/QVGen-CogVideoX-2B-W4A4",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "Harahan/QVGen-CogVideoX-2B-W4A4",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W36"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W36"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-05T11:13:13.736Z",
          "last_at": "2026-09-05T11:13:13.736Z",
          "caps": [],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "Interchained/Muse-Spark-Open-816B-A42B-GGUF:Q4_K_M",
      "family": "muse-spark",
      "member": "816b-a42b",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "meta-models/Muse-Spark-Open-816B-A42B",
      "hub_repo": "meta-models/Muse-Spark-Open-816B-A42B",
      "publisher": "Interchained",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 476446,
      "active_share": 0.0515,
      "licence": {
        "id": "other",
        "name": "muse-model-license",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-models/Muse-Spark-Open-816B-A42B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta/muse-spark-open",
        "ids": [
          "meta/muse-spark-open"
        ],
        "usd_per_m_in": 1.25,
        "usd_per_m_out": 4.25,
        "usd_per_m_cached_in": 0.15,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "Jundot/DeepSeek-V4-Flash-0731-oQ4e-mtp",
      "family": "deepseek-v4",
      "member": "flash-0731~jundot-oq4e-mtp",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4-Flash-0731",
      "hub_repo": "deepseek-ai/DeepSeek-V4-Flash-0731",
      "publisher": "Jundot",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 158146,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4-Flash-0731",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4-flash-0731",
        "ids": [
          "deepseek/deepseek-v4-flash-0731",
          "deepseek/deepseek-v4-flash-0731:batch"
        ],
        "usd_per_m_in": 0.0434,
        "usd_per_m_out": 0.3,
        "usd_per_m_cached_in": 0.035,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.144,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "Jundot/GLM-5.3-Flash-oQ4e-mtp",
      "family": "glm-5-3",
      "member": "flash~jundot-oq4e-mtp",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.3-Flash",
      "hub_repo": "zai-org/GLM-5.3-Flash",
      "publisher": "Jundot",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 177767,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.3-Flash",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.3-flash",
        "ids": [
          "z-ai/glm-5.3-flash",
          "z-ai/glm-5.3-flash:batch"
        ],
        "usd_per_m_in": 0.0399,
        "usd_per_m_out": 0.0784,
        "usd_per_m_cached_in": 0.03,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.18,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "Jundot/Qwen3.6-35B-A3B-oQ4e-mtp",
      "family": "qwen-3-6",
      "member": "35b-a3b~jundot-oq4e-mtp",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3.6-35B-A3B",
      "hub_repo": "Qwen/Qwen3.6-35B-A3B",
      "publisher": "Jundot",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 21108,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.6-35B-A3B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.6-35b-a3b",
        "ids": [
          "qwen/qwen3.6-35b-a3b"
        ],
        "usd_per_m_in": 0.1,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.025,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "Jundot/Qwen3.8-27B-oQ4e-mtp",
      "family": "qwen-3-8",
      "member": "27b~jundot-oq4e-mtp",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "Jundot",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 16705,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "Jundot/Qwen3.8-Flash-Next-oQ4e-mtp",
      "family": "qwen-3-8",
      "member": "flash-next~jundot-oq4e-mtp",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3.8-Flash-Next",
      "hub_repo": "Qwen/Qwen3.8-Flash-Next",
      "publisher": "Jundot",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 102682,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "qwen-community-1.0",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-Flash-Next",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.8-flash",
        "ids": [
          "qwen/qwen3.8-flash"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.47,
        "usd_per_m_cached_in": 0.016,
        "context_length": 1000000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "LanguaMan/stable-audio-open-1-0-model",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "LanguaMan/stable-audio-open-1-0-model",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-30T01:41:15.832Z",
          "last_at": "2026-08-30T01:41:15.832Z",
          "caps": [],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "Lightricks/LTX-Video-0.9.5",
      "family": "ltx-video",
      "member": "2b-0-9-5",
      "quant": "BF16",
      "engine": "diffusers",
      "modality": "image",
      "base": null,
      "hub_repo": "Lightricks/LTX-Video-0.9.5",
      "publisher": "Lightricks",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 24215,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Lightricks/LTX-Video-0.9.5",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T06:38:05.516Z",
          "last_at": "2026-10-21T06:38:05.516Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "Lightricks/LTX-Video-ICLoRA-detailer-13b-0.9.8",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "Lightricks/LTX-Video-ICLoRA-detailer-13b-0.9.8",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W36"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W36"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-05T11:11:36.950Z",
          "last_at": "2026-09-05T11:11:36.950Z",
          "caps": [],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "LiquidAI/LFM2.5-24B-A2B-FP8",
      "family": "lfm-2-5",
      "member": "24b-a2b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "LiquidAI/LFM2.5-24B-A2B",
      "hub_repo": "LiquidAI/LFM2.5-24B-A2B",
      "publisher": "LiquidAI",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 23600,
      "active_share": 0.0833,
      "licence": {
        "id": "other",
        "name": "lfm1.0",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/LiquidAI/LFM2.5-24B-A2B",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "liquid/lfm-2.5-24b-a2b",
        "ids": [
          "liquid/lfm-2.5-24b-a2b"
        ],
        "usd_per_m_in": 0.05,
        "usd_per_m_out": 0.7,
        "usd_per_m_cached_in": 0.025,
        "context_length": 128000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 256.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5098,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "LiquidAI/LFM2.5-24B-A2B-MLX-4bit",
      "family": "lfm-2-5",
      "member": "24b-a2b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "LiquidAI/LFM2.5-24B-A2B",
      "hub_repo": "LiquidAI/LFM2.5-24B-A2B",
      "publisher": "LiquidAI",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 12938,
      "active_share": 0.0833,
      "licence": {
        "id": "other",
        "name": "lfm1.0",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/LiquidAI/LFM2.5-24B-A2B",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "liquid/lfm-2.5-24b-a2b",
        "ids": [
          "liquid/lfm-2.5-24b-a2b"
        ],
        "usd_per_m_in": 0.05,
        "usd_per_m_out": 0.7,
        "usd_per_m_cached_in": 0.025,
        "context_length": 128000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "MiniMaxAI/MiniMax-M2.5",
      "family": "minimax-m2-5",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.5",
      "hub_repo": "MiniMaxAI/MiniMax-M2.5",
      "publisher": "MiniMaxAI",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 219473,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.5",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.5",
        "ids": [
          "minimax/minimax-m2.5"
        ],
        "usd_per_m_in": 0.27,
        "usd_per_m_out": 0.95,
        "usd_per_m_cached_in": 0.03,
        "context_length": 198000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 126.1458,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "MiniMaxAI/MiniMax-M2.7",
      "family": "minimax-m2-7",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.7",
      "hub_repo": "MiniMaxAI/MiniMax-M2.7",
      "publisher": "MiniMaxAI",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 219473,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "other",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.7",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.7",
        "ids": [
          "minimax/minimax-m2.7"
        ],
        "usd_per_m_in": 0.21,
        "usd_per_m_out": 0.84,
        "usd_per_m_cached_in": 0.042,
        "context_length": 196608,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 126.1458,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "PrunaAI/Pruna-Qwen-Image-2.1",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "PrunaAI/Pruna-Qwen-Image-2.1",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W40",
        "2026-W42"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W40",
            "2026-W42"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-01T06:56:47.231Z",
          "last_at": "2026-10-16T07:34:24.723Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "QwQbb/Hy-MT2-30B-A3B-MLX-4bit",
      "family": "hy-mt-2",
      "member": "30b-a3b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "tencent/Hy-MT2-30B-A3B",
      "hub_repo": "tencent/Hy-MT2-30B-A3B",
      "publisher": "QwQbb",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 16134,
      "active_share": 0.1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy-MT2-30B-A3B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy-mt2-30b-a3b",
        "ids": [
          "tencent/hy-mt2-30b-a3b"
        ],
        "usd_per_m_in": 0.074,
        "usd_per_m_out": 0.295,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "Qwen/Qwen3-1.7B",
      "family": "qwen-3",
      "member": "1.7b",
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "Qwen/Qwen3-1.7B",
      "publisher": "qwen",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-1.7B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 1.223,
          "first_at": "2026-10-23T19:52:27.499Z",
          "last_at": "2026-10-23T19:52:27.499Z",
          "caps": [
            "code.reason",
            "faithful_tool_result",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "Qwen/Qwen3-14B-AWQ",
      "family": "qwen-3",
      "member": "14b",
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "Qwen/Qwen3-14B-AWQ",
      "publisher": "qwen",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-14B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3-14b",
        "ids": [
          "qwen/qwen3-14b"
        ],
        "usd_per_m_in": 0.12,
        "usd_per_m_out": 0.24,
        "usd_per_m_cached_in": null,
        "context_length": 40960,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 2.965,
          "first_at": "2026-10-23T18:24:45.561Z",
          "last_at": "2026-10-23T18:24:45.561Z",
          "caps": [
            "code.reason",
            "faithful_tool_result",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.ml",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "Qwen/Qwen3-4B",
      "family": "qwen-3",
      "member": "4b",
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "Qwen/Qwen3-4B",
      "publisher": "qwen",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-4B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 2.796,
          "first_at": "2026-10-20T10:09:23.154Z",
          "last_at": "2026-10-20T10:09:23.154Z",
          "caps": [
            "code.reason",
            "faithful_tool_result",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "Qwen/Qwen3-8B",
      "family": "qwen-3",
      "member": "8b",
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "Qwen/Qwen3-8B",
      "publisher": "qwen",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-8B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 4.978,
          "first_at": "2026-10-23T20:45:46.870Z",
          "last_at": "2026-10-23T20:45:46.870Z",
          "caps": [
            "code.reason",
            "faithful_tool_result",
            "lang.bn",
            "lang.en",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "Qwen/Qwen3-Coder-Next-FP8",
      "family": "qwen-3",
      "member": "coder-next",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "Qwen/Qwen3-Coder-Next-FP8",
      "hub_repo": "Qwen/Qwen3-Coder-Next-FP8",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 76657,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-Coder-Next-FP8",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3-coder-next",
        "ids": [
          "qwen/qwen3-coder-next"
        ],
        "usd_per_m_in": 0.12,
        "usd_per_m_out": 0.8,
        "usd_per_m_cached_in": 0.07,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.4303,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "Qwen/Qwen3.6-27B-FP8",
      "family": "qwen-3-6",
      "member": "27b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "Qwen/Qwen3.6-27B",
      "hub_repo": "Qwen/Qwen3.6-27B",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 29436,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.6-27B",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.6-27b",
        "ids": [
          "qwen/qwen3.6-27b"
        ],
        "usd_per_m_in": 0.32,
        "usd_per_m_out": 3.2,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.9091,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "Qwen/Qwen3.6-35B-A3B-FP8",
      "family": "qwen-3-6",
      "member": "35b-a3b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "Qwen/Qwen3.6-35B-A3B",
      "hub_repo": "Qwen/Qwen3.6-35B-A3B",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 35728,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.6-35B-A3B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.6-35b-a3b",
        "ids": [
          "qwen/qwen3.6-35b-a3b"
        ],
        "usd_per_m_in": 0.1,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.025,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 164.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8537,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "Qwen/Qwen3.8-27B-FP8",
      "family": "qwen-3-8",
      "member": "27b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 29436,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.9091,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "Qwen/Qwen3.8-Flash-Next-FP8",
      "family": "qwen-3-8",
      "member": "flash-next",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "Qwen/Qwen3.8-Flash-Next",
      "hub_repo": "Qwen/Qwen3.8-Flash-Next",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 176928,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "qwen-community-1.0",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-Flash-Next",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.8-flash",
        "ids": [
          "qwen/qwen3.8-flash"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.47,
        "usd_per_m_cached_in": 0.016,
        "context_length": 1000000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 89.6534,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "RedHatAI/Kimi-K2.6-FP8-BLOCK",
      "family": "kimi-k2-6",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.6",
      "hub_repo": "moonshotai/Kimi-K2.6",
      "publisher": "RedHatAI",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 983174,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.6",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.6",
        "ids": [
          "moonshotai/kimi-k2.6"
        ],
        "usd_per_m_in": 0.65,
        "usd_per_m_out": 3.41,
        "usd_per_m_cached_in": 0.15,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1787.5232,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "RedHatAI/Llama-3.2-1B-Instruct-FP8-dynamic",
      "family": "llama-3-2",
      "member": "1b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-1B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-1B-Instruct",
      "publisher": "RedHatAI",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 1930,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-1B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-1b-instruct",
        "ids": [
          "meta-llama/llama-3.2-1b-instruct"
        ],
        "usd_per_m_in": 0.027,
        "usd_per_m_out": 0.201,
        "usd_per_m_cached_in": null,
        "context_length": 60000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 261.32,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4328,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "RedHatAI/Llama-3.2-3B-Instruct-FP8-dynamic",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-3B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-3B-Instruct",
      "publisher": "RedHatAI",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 4192,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 120.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9554,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "RedHatAI/Muse-Glimmer-30B-FP8-block",
      "family": "muse-glimmer",
      "member": "30b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "meta-models/Muse-Glimmer-30B",
      "hub_repo": "meta-models/Muse-Glimmer-30B",
      "publisher": "RedHatAI",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 32799,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/meta-models/Muse-Glimmer-30B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta/muse-glimmer-30b",
        "ids": [
          "meta/muse-glimmer-30b",
          "meta/muse-glimmer-30b:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.04,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.96,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.987,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "RedHatAI/Qwen3.5-9B-FP8-dynamic",
      "family": "qwen-3-5",
      "member": "9b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "Qwen/Qwen3.5-9B",
      "hub_repo": "Qwen/Qwen3.5-9B",
      "publisher": "RedHatAI",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 13357,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.5-9B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.5-9b",
        "ids": [
          "qwen/qwen3.5-9b"
        ],
        "usd_per_m_in": 0.037,
        "usd_per_m_out": 0.054,
        "usd_per_m_cached_in": 0.033,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 37.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.2415,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "RedHatAI/gemma-4-26B-A4B-it-FP8-dynamic",
      "family": "gemma-4",
      "member": "26b-a4b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "google/gemma-4-26B-A4B-it",
      "hub_repo": "google/gemma-4-26B-A4B-it",
      "publisher": "RedHatAI",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 27314,
      "active_share": 0.1538,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-26B-A4B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "google/gemma-4-26b-a4b-it",
        "ids": [
          "google/gemma-4-26b-a4b-it",
          "google/gemma-4-26b-a4b-it:free"
        ],
        "usd_per_m_in": 0.0675,
        "usd_per_m_out": 0.225,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 262144,
        "free_variant": true,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 120.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1144,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "RedHatAI/phi-4-FP8-dynamic",
      "family": "phi-4",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "microsoft/phi-4",
      "hub_repo": "microsoft/phi-4",
      "publisher": "RedHatAI",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 14964,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/phi-4",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "microsoft/phi-4",
        "ids": [
          "microsoft/phi-4"
        ],
        "usd_per_m_in": 0.07,
        "usd_per_m_out": 0.14,
        "usd_per_m_cached_in": null,
        "context_length": 16384,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 33.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.6708,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "RedHatAI/phi-4-quantized.w8a8",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "RedHatAI/phi-4-quantized.w8a8",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 5.471,
          "first_at": "2026-10-23T21:03:11.717Z",
          "last_at": "2026-10-23T21:03:11.717Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.mr",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "SPRINGLab/F5-Hindi-24KHz",
      "family": "f5-tts",
      "member": "hindi-24khz",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "SWivid/F5-TTS",
      "hub_repo": "SWivid/F5-TTS",
      "publisher": "SPRINGLab",
      "derivative": false,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 604,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/SWivid/F5-TTS",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T21:48:42.302Z",
          "last_at": "2026-10-27T12:21:30.056Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:19:52.134Z",
          "last_at": "2026-11-07T02:44:42.274Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-28T18:25:41.676Z",
          "last_at": "2026-11-14T18:57:18.035Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "SWivid/F5-TTS",
      "family": "f5-tts",
      "member": "v1-base",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "SWivid/F5-TTS",
      "hub_repo": "SWivid/F5-TTS",
      "publisher": "SWivid",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 6430,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/SWivid/F5-TTS",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:19:24.553Z",
          "last_at": "2026-11-07T02:44:12.556Z",
          "caps": [
            "audio",
            "audio.out",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "Spangler3000/MiniMax-M3-Mixed-4.5bit-MLX",
      "family": "minimax-m3",
      "member": "base~spangler3000-mixed-4",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M3",
      "hub_repo": "MiniMaxAI/MiniMax-M3",
      "publisher": "Spangler3000",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 257788,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "minimax-community",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M3",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m3",
        "ids": [
          "minimax/minimax-m3",
          "minimax/minimax-m3:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.06,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.88,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "THUDM/CogVideoX-2b",
      "family": "cogvideox",
      "member": "2b",
      "quant": "BF16",
      "engine": "diffusers",
      "modality": "image",
      "base": null,
      "hub_repo": "THUDM/CogVideoX-2b",
      "publisher": "THUDM",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 13137,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/THUDM/CogVideoX-2b",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [
        "2026-W38",
        "2026-W43",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T06:35:45.949Z",
          "last_at": "2026-10-21T06:35:45.949Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-06T12:09:40.542Z",
          "last_at": "2026-11-06T12:09:40.542Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-17T01:06:35.345Z",
          "last_at": "2026-09-17T01:06:35.345Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T13:03:41.444Z",
          "last_at": "2026-09-15T13:03:41.444Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "TensorFold/DeepSeek-V4-Flash-0731-MXFP4-MLX",
      "family": "deepseek-v4",
      "member": "flash-0731",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4-Flash-0731",
      "hub_repo": "deepseek-ai/DeepSeek-V4-Flash-0731",
      "publisher": "TensorFold",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 159342,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4-Flash-0731",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4-flash-0731",
        "ids": [
          "deepseek/deepseek-v4-flash-0731",
          "deepseek/deepseek-v4-flash-0731:batch"
        ],
        "usd_per_m_in": 0.0434,
        "usd_per_m_out": 0.3,
        "usd_per_m_cached_in": 0.035,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.144,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "TensorFold/GLM-5.3-Flash-MLX-4bit-MTP",
      "family": "glm-5-3",
      "member": "flash~tensorfold-mtp",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.3-Flash",
      "hub_repo": "zai-org/GLM-5.3-Flash",
      "publisher": "TensorFold",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 173291,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.3-Flash",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.3-flash",
        "ids": [
          "z-ai/glm-5.3-flash",
          "z-ai/glm-5.3-flash:batch"
        ],
        "usd_per_m_in": 0.0399,
        "usd_per_m_out": 0.0784,
        "usd_per_m_cached_in": 0.03,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.18,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "TensorFold/GLM-5.3-Flash-MLX-oQ4-MTP",
      "family": "glm-5-3",
      "member": "flash~tensorfold-oq4-mtp",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.3-Flash",
      "hub_repo": "zai-org/GLM-5.3-Flash",
      "publisher": "TensorFold",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 177143,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.3-Flash",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.3-flash",
        "ids": [
          "z-ai/glm-5.3-flash",
          "z-ai/glm-5.3-flash:batch"
        ],
        "usd_per_m_in": 0.0399,
        "usd_per_m_out": 0.0784,
        "usd_per_m_cached_in": 0.03,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.18,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "TensorFold/MiMo-V2.6-Flash-RL-MLX-4bit-MTP",
      "family": "mimo-v2-6",
      "member": "flash-rl~tensorfold-mtp",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "hub_repo": "XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "publisher": "TensorFold",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 163689,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.6-Flash-RL",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.6-flash",
        "ids": [
          "xiaomi/mimo-v2.6-flash"
        ],
        "usd_per_m_in": 0.14,
        "usd_per_m_out": 0.28,
        "usd_per_m_cached_in": 0.0028,
        "context_length": 1050000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "TensorFold/Muse-Glimmer-30B-oQ4",
      "family": "muse-glimmer",
      "member": "30b~tensorfold-oq4",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "meta-models/Muse-Glimmer-30B",
      "hub_repo": "meta-models/Muse-Glimmer-30B",
      "publisher": "TensorFold",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 19317,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/meta-models/Muse-Glimmer-30B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta/muse-glimmer-30b",
        "ids": [
          "meta/muse-glimmer-30b",
          "meta/muse-glimmer-30b:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.04,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.96,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "TitanML/GLM-5.1-FP8",
      "family": "glm-5-1",
      "member": "base~titanml",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": null,
      "hub_repo": "TitanML/GLM-5.1-FP8",
      "publisher": "TitanML",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 721147,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/TitanML/GLM-5.1-FP8",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.1",
        "ids": [
          "z-ai/glm-5.1"
        ],
        "usd_per_m_in": 1.4,
        "usd_per_m_out": 4.4,
        "usd_per_m_cached_in": 0.26,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 997.7124,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "True2456/Step-3.7-173B-REAM-4.6bit-MLX",
      "family": "step-3-7",
      "member": "flash~true2456-ream-4",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "stepfun-ai/Step-3.7-Flash",
      "hub_repo": "stepfun-ai/Step-3.7-Flash",
      "publisher": "True2456",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 96497,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "stepfun/step-3.7-flash",
        "ids": [
          "stepfun/step-3.7-flash"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 1.15,
        "usd_per_m_cached_in": 0.04,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "True2456/Step-3.7-173B-REAP-4.6bit-MLX",
      "family": "step-3-7",
      "member": "flash~true2456-reap-4",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "stepfun-ai/Step-3.7-Flash",
      "hub_repo": "stepfun-ai/Step-3.7-Flash",
      "publisher": "True2456",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 95194,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "stepfun/step-3.7-flash",
        "ids": [
          "stepfun/step-3.7-flash"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 1.15,
        "usd_per_m_cached_in": 0.04,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "UsefulSensors/moonshine-base",
      "family": "moonshine",
      "member": "base",
      "quant": null,
      "engine": "speech-pod",
      "modality": "speech",
      "base": null,
      "hub_repo": "UsefulSensors/moonshine-base",
      "publisher": "UsefulSensors",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 234,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/UsefulSensors/moonshine-base",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T04:54:22.429Z",
          "last_at": "2026-11-14T13:21:38.168Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-20T09:29:43.546Z",
          "last_at": "2026-10-20T09:29:43.546Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:16:56.943Z",
          "last_at": "2026-11-03T22:16:56.943Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "UsefulSensors/moonshine-tiny",
      "family": "moonshine",
      "member": "tiny",
      "quant": null,
      "engine": "speech-pod",
      "modality": "speech",
      "base": null,
      "hub_repo": "UsefulSensors/moonshine-tiny",
      "publisher": "UsefulSensors",
      "derivative": false,
      "cohort": 2,
      "rung": 9,
      "disk_mib": 103,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/UsefulSensors/moonshine-tiny",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W30",
        "2026-W37",
        "2026-W39",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T04:50:38.161Z",
          "last_at": "2026-11-12T18:32:28.736Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-19T21:12:26.660Z",
          "last_at": "2026-10-19T21:17:02.974Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:17:23.023Z",
          "last_at": "2026-11-03T22:17:23.023Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W30"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-07-26T16:01:37.829Z",
          "last_at": "2026-07-26T16:01:37.829Z",
          "caps": [
            "audio"
          ],
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37",
            "2026-W39"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-13T07:05:16.550Z",
          "last_at": "2026-09-21T08:26:20.102Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "Wan-AI/Wan2.1-T2V-1.3B",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "Wan-AI/Wan2.1-T2V-1.3B",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W36",
        "2026-W37",
        "2026-W38",
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T06:32:57.953Z",
          "last_at": "2026-10-21T06:32:57.953Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-13T16:11:32.081Z",
          "last_at": "2026-09-13T16:11:32.081Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W36"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-05T11:17:39.818Z",
          "last_at": "2026-09-05T11:17:39.818Z",
          "caps": [],
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-14T05:50:54.814Z",
          "last_at": "2026-09-14T05:50:54.814Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "Wan-AI/Wan2.1-T2V-1.3B-Diffusers",
      "family": "wan-2",
      "member": "t2v-1.3b",
      "quant": null,
      "engine": "diffusers",
      "modality": "image",
      "base": null,
      "hub_repo": "Wan-AI/Wan2.1-T2V-1.3B-Diffusers",
      "publisher": "Wan-AI",
      "derivative": false,
      "cohort": 2,
      "rung": 9,
      "disk_mib": 27568,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Wan-AI/Wan2.1-T2V-1.3B-Diffusers",
        "fetched_at": "2026-11-10T15:57:15.731Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T06:40:45.664Z",
          "last_at": "2026-10-22T08:04:30.830Z",
          "caps": [
            "video.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "XiaomiMiMo/MiMo-V2.5",
      "family": "mimo-v2-5",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "XiaomiMiMo/MiMo-V2.5",
      "hub_repo": "XiaomiMiMo/MiMo-V2.5",
      "publisher": "XiaomiMiMo",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 301068,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.5",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.5",
        "ids": [
          "xiaomi/mimo-v2.5"
        ],
        "usd_per_m_in": 0.4,
        "usd_per_m_out": 2,
        "usd_per_m_cached_in": 0.08,
        "context_length": 1000000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 212.2247,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "XiaomiMiMo/MiMo-V2.5-Pro",
      "family": "mimo-v2-5",
      "member": "pro",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "XiaomiMiMo/MiMo-V2.5-Pro",
      "hub_repo": "XiaomiMiMo/MiMo-V2.5-Pro",
      "publisher": "XiaomiMiMo",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 985517,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.5-Pro",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.5-pro",
        "ids": [
          "xiaomi/mimo-v2.5-pro"
        ],
        "usd_per_m_in": 0.435,
        "usd_per_m_out": 0.87,
        "usd_per_m_cached_in": 0.0036,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1791.2619,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "family": "mimo-v2-6",
      "member": "flash-rl",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "hub_repo": "XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "publisher": "XiaomiMiMo",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 169507,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.6-Flash-RL",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.6-flash",
        "ids": [
          "xiaomi/mimo-v2.6-flash"
        ],
        "usd_per_m_in": 0.14,
        "usd_per_m_out": 0.28,
        "usd_per_m_cached_in": 0.0028,
        "context_length": 1050000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 83.7157,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "XiaomiMiMo/MiMo-V2.6-Pro-RL",
      "family": "mimo-v2-6",
      "member": "pro-rl",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "XiaomiMiMo/MiMo-V2.6-Pro-RL",
      "hub_repo": "XiaomiMiMo/MiMo-V2.6-Pro-RL",
      "publisher": "XiaomiMiMo",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 546891,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.6-Pro-RL",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.6-pro",
        "ids": [
          "xiaomi/mimo-v2.6-pro"
        ],
        "usd_per_m_in": 0.435,
        "usd_per_m_out": 0.87,
        "usd_per_m_cached_in": 0.0036,
        "context_length": 1050000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 604.9878,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "Youssofal/MiniMax-M2.7-Abliterated-Heretic-GGUF:Q4_K_M",
      "family": "minimax-m2-7",
      "member": "base~youssofal-abliterated-heretic",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.7",
      "hub_repo": "MiniMaxAI/MiniMax-M2.7",
      "publisher": "Youssofal",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 131933,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "other",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.7",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.7",
        "ids": [
          "minimax/minimax-m2.7"
        ],
        "usd_per_m_in": 0.21,
        "usd_per_m_out": 0.84,
        "usd_per_m_cached_in": 0.042,
        "context_length": 196608,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "Youssofal/Qwen3.8-27B-MTPLX-Optimized-Speed",
      "family": "qwen-3-8",
      "member": "27b~youssofal-mtplx-optimized-speed",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "Youssofal",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 19725,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "Youssofal/Qwen3.8-Flash-Next-MTPLX-Optimized-Speed",
      "family": "qwen-3-8",
      "member": "flash-next~youssofal-mtplx-optimized-speed",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3.8-Flash-Next",
      "hub_repo": "Qwen/Qwen3.8-Flash-Next",
      "publisher": "Youssofal",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 109708,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "qwen-community-1.0",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-Flash-Next",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.8-flash",
        "ids": [
          "qwen/qwen3.8-flash"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.47,
        "usd_per_m_cached_in": 0.016,
        "context_length": 1000000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "aidiffuser/Kimi-K2.6-MLX-4bit",
      "family": "kimi-k2-6",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.6",
      "hub_repo": "moonshotai/Kimi-K2.6",
      "publisher": "aidiffuser",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 627158,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.6",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.6",
        "ids": [
          "moonshotai/kimi-k2.6"
        ],
        "usd_per_m_in": 0.65,
        "usd_per_m_out": 3.41,
        "usd_per_m_cached_in": 0.15,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "aidsoid/Hy-MT2-1.8B-MLX-4bit",
      "family": "hy-mt-2",
      "member": "1.8b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "tencent/Hy-MT2-1.8B",
      "hub_repo": "tencent/Hy-MT2-1.8B",
      "publisher": "aidsoid",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 961,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy-MT2-1.8B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy-mt2-1.8b",
        "ids": [
          "tencent/hy-mt2-1.8b"
        ],
        "usd_per_m_in": 0.044,
        "usd_per_m_out": 0.177,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "alibaba-pai/Wan2.1-Fun-V1.1-1.3B-InP",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "alibaba-pai/Wan2.1-Fun-V1.1-1.3B-InP",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W40",
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W40"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-01T07:59:54.117Z",
          "last_at": "2026-10-01T07:59:54.117Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T06:31:47.409Z",
          "last_at": "2026-10-21T06:31:47.409Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "autospare/Mistral-Small-4-119B-2603",
      "family": "mistral-small-4",
      "member": "119b-2603~autospare",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": null,
      "hub_repo": "autospare/Mistral-Small-4-119B-2603",
      "publisher": "autospare",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 230646,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/autospare/Mistral-Small-4-119B-2603",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-small-2603",
        "ids": [
          "mistralai/mistral-small-2603"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": 0.015,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 136.6338,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "avlp12/GLM-5.2-Alis-MLX-Dynamic-3.5bpw",
      "family": "glm-5-2",
      "member": "base~avlp12-alis-dynamic-3-5bpw",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.2",
      "hub_repo": "zai-org/GLM-5.2",
      "publisher": "avlp12",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 317571,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.2",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.2",
        "ids": [
          "z-ai/glm-5.2"
        ],
        "usd_per_m_in": 0.007,
        "usd_per_m_out": 6.8,
        "usd_per_m_cached_in": 0.006,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "avlp12/Hy3-Alis-MLX-Dynamic",
      "family": "hy-3",
      "member": "base~avlp12-alis-dynamic",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "tencent/Hy3",
      "hub_repo": "tencent/Hy3",
      "publisher": "avlp12",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 160661,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy3",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy3",
        "ids": [
          "tencent/hy3"
        ],
        "usd_per_m_in": 0.0825,
        "usd_per_m_out": 0.33,
        "usd_per_m_cached_in": 0.020625,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "axteam/GLM-OCR-FP8-dynamic",
      "family": "glm-ocr",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-OCR",
      "hub_repo": "zai-org/GLM-OCR",
      "publisher": "axteam",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 2060,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-OCR",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 1.458,
          "first_at": "2026-11-07T01:45:29.834Z",
          "last_at": "2026-11-07T01:45:29.834Z",
          "caps": [
            "code.reason",
            "code.write",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "est_tokens_per_sec": 244.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4624,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "baa-ai/DeepSeek-V3.2-RAM-350GB-MLX",
      "family": "deepseek-v3-2",
      "member": "base~baa-ai-ram-350gb",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V3.2",
      "hub_repo": "deepseek-ai/DeepSeek-V3.2",
      "publisher": "baa-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 358882,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.2",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v3.2",
        "ids": [
          "deepseek/deepseek-v3.2"
        ],
        "usd_per_m_in": 0.26,
        "usd_per_m_out": 0.38,
        "usd_per_m_cached_in": 0.13,
        "context_length": 163840,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "baa-ai/GLM-4.7-Flash-RAM-20GB-MLX",
      "family": "glm-4-7",
      "member": "flash~baa-ai-ram-20gb",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-4.7-Flash",
      "hub_repo": "zai-org/GLM-4.7-Flash",
      "publisher": "baa-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 18376,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.7-Flash",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.7-flash",
        "ids": [
          "z-ai/glm-4.7-flash"
        ],
        "usd_per_m_in": 0.0605,
        "usd_per_m_out": 0.4,
        "usd_per_m_cached_in": null,
        "context_length": 200000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "baa-ai/GLM-5.1-RAM-270GB-MLX",
      "family": "glm-5-1",
      "member": "base~baa-ai-ram-270gb",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.1",
      "hub_repo": "zai-org/GLM-5.1",
      "publisher": "baa-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 275766,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.1",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.1",
        "ids": [
          "z-ai/glm-5.1"
        ],
        "usd_per_m_in": 1.4,
        "usd_per_m_out": 4.4,
        "usd_per_m_cached_in": 0.26,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "baa-ai/GLM-5.1-RAM-420GB-MLX",
      "family": "glm-5-1",
      "member": "base~baa-ai-ram-420gb",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.1",
      "hub_repo": "zai-org/GLM-5.1",
      "publisher": "baa-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 364137,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.1",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.1",
        "ids": [
          "z-ai/glm-5.1"
        ],
        "usd_per_m_in": 1.4,
        "usd_per_m_out": 4.4,
        "usd_per_m_cached_in": 0.26,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "baa-ai/MiniMax-M2.7-RAM-100GB-MLX",
      "family": "minimax-m2-7",
      "member": "base~baa-ai-ram-100gb",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.7",
      "hub_repo": "MiniMaxAI/MiniMax-M2.7",
      "publisher": "baa-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 95464,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "other",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.7",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.7",
        "ids": [
          "minimax/minimax-m2.7"
        ],
        "usd_per_m_in": 0.21,
        "usd_per_m_out": 0.84,
        "usd_per_m_cached_in": 0.042,
        "context_length": 196608,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "baa-ai/MiniMax-M2.7-RAM-120GB-MLX",
      "family": "minimax-m2-7",
      "member": "base~baa-ai-ram-120gb",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.7",
      "hub_repo": "MiniMaxAI/MiniMax-M2.7",
      "publisher": "baa-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 114543,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "other",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.7",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.7",
        "ids": [
          "minimax/minimax-m2.7"
        ],
        "usd_per_m_in": 0.21,
        "usd_per_m_out": 0.84,
        "usd_per_m_cached_in": 0.042,
        "context_length": 196608,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "baa-ai/MiniMax-M2.7-RAM-91GB-MLX",
      "family": "minimax-m2-7",
      "member": "base~baa-ai-ram-91gb",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.7",
      "hub_repo": "MiniMaxAI/MiniMax-M2.7",
      "publisher": "baa-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 91959,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "other",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.7",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.7",
        "ids": [
          "minimax/minimax-m2.7"
        ],
        "usd_per_m_in": 0.21,
        "usd_per_m_out": 0.84,
        "usd_per_m_cached_in": 0.042,
        "context_length": 196608,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "bartowski/Hy3-GGUF:Q4_K_M",
      "family": "hy-3",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "tencent/Hy3",
      "hub_repo": "tencent/Hy3",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 173719,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy3",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy3",
        "ids": [
          "tencent/hy3"
        ],
        "usd_per_m_in": 0.0825,
        "usd_per_m_out": 0.33,
        "usd_per_m_cached_in": 0.020625,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "bartowski/MiniMaxAI_MiniMax-M2.7-GGUF:Q4_K_M",
      "family": "minimax-m2-7",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.7",
      "hub_repo": "MiniMaxAI/MiniMax-M2.7",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 132375,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "other",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.7",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.7",
        "ids": [
          "minimax/minimax-m2.7"
        ],
        "usd_per_m_in": 0.21,
        "usd_per_m_out": 0.84,
        "usd_per_m_cached_in": 0.042,
        "context_length": 196608,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "bartowski/deepseek-ai_DeepSeek-V3.1-GGUF:Q4_K_M",
      "family": "deepseek-v3-1",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V3.1",
      "hub_repo": "deepseek-ai/DeepSeek-V3.1",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 390269,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.1",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-chat-v3.1",
        "ids": [
          "deepseek/deepseek-chat-v3.1"
        ],
        "usd_per_m_in": 0.25,
        "usd_per_m_out": 0.95,
        "usd_per_m_cached_in": 0.13,
        "context_length": 163840,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "bb1975/MiMo-V2.5-Pro",
      "family": "mimo-v2-5",
      "member": "pro~bb1975",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": null,
      "hub_repo": "bb1975/MiMo-V2.5-Pro",
      "publisher": "bb1975",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 985517,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/bb1975/MiMo-V2.5-Pro",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.5-pro",
        "ids": [
          "xiaomi/mimo-v2.5-pro"
        ],
        "usd_per_m_in": 0.435,
        "usd_per_m_out": 0.87,
        "usd_per_m_cached_in": 0.0036,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1791.2619,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "catalystsec/GLM-5.1-4bit",
      "family": "glm-5-1",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.1",
      "hub_repo": "zai-org/GLM-5.1",
      "publisher": "catalystsec",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 399228,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.1",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.1",
        "ids": [
          "z-ai/glm-5.1"
        ],
        "usd_per_m_in": 1.4,
        "usd_per_m_out": 4.4,
        "usd_per_m_cached_in": 0.26,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "catalystsec/Kimi-K3-Audit-240E-MLX-MXFP4",
      "family": "kimi-k3",
      "member": "base~catalystsec-audit-240e",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K3",
      "hub_repo": "moonshotai/Kimi-K3",
      "publisher": "catalystsec",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 417677,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "kimi-k3",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K3",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k3",
        "ids": [
          "moonshotai/kimi-k3",
          "moonshotai/kimi-k3:batch"
        ],
        "usd_per_m_in": 0.192,
        "usd_per_m_out": 16,
        "usd_per_m_cached_in": 0.182,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 11.4,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "catalystsec/Kimi-K3-Audit-451E-MLX-MXFP4",
      "family": "kimi-k3",
      "member": "base~catalystsec-audit-451e",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K3",
      "hub_repo": "moonshotai/Kimi-K3",
      "publisher": "catalystsec",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 742666,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "kimi-k3",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K3",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k3",
        "ids": [
          "moonshotai/kimi-k3",
          "moonshotai/kimi-k3:batch"
        ],
        "usd_per_m_in": 0.192,
        "usd_per_m_out": 16,
        "usd_per_m_cached_in": 0.182,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 11.4,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "claude-haiku-4-5",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": null,
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W35",
        "2026-W37",
        "2026-W38",
        "2026-W39",
        "2026-W40",
        "2026-W41",
        "2026-W42",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 3,
          "seasons": [
            "2026-W35",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": 93.24,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-24T06:02:38.410Z",
          "last_at": "2026-11-14T10:56:00.687Z",
          "caps": [
            "tools"
          ],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "claude-opus-5",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": null,
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W35",
        "2026-W37",
        "2026-W38",
        "2026-W39",
        "2026-W40",
        "2026-W41",
        "2026-W42",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 3,
          "seasons": [
            "2026-W35",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": 51.53,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-24T06:02:13.071Z",
          "last_at": "2026-11-14T10:54:46.365Z",
          "caps": [
            "tools"
          ],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "claude-sonnet-5",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": null,
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W35",
        "2026-W37",
        "2026-W38",
        "2026-W39",
        "2026-W40",
        "2026-W41",
        "2026-W42",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 3,
          "seasons": [
            "2026-W35",
            "2026-W37",
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": 95.1,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-24T06:02:23.338Z",
          "last_at": "2026-11-14T10:55:10.132Z",
          "caps": [
            "tools"
          ],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "coqui/XTTS-v2",
      "family": "xtts",
      "member": "v2",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "coqui/XTTS-v2",
      "hub_repo": "coqui/XTTS-v2",
      "publisher": "coqui",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 1989,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "coqui-public-model-license",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/coqui/XTTS-v2",
        "fetched_at": "2026-11-10T15:57:15.731Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T21:26:47.093Z",
          "last_at": "2026-11-06T17:14:33.298Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "dareallest207/Muse-Spark-Open-816B-A42B-oQ4e",
      "family": "muse-spark",
      "member": "816b-a42b~dareallest207-open-oq4e",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "meta-models/Muse-Spark-Open-816B-A42B",
      "hub_repo": "meta-models/Muse-Spark-Open-816B-A42B",
      "publisher": "dareallest207",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 453840,
      "active_share": 0.0515,
      "licence": {
        "id": "other",
        "name": "muse-model-license",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-models/Muse-Spark-Open-816B-A42B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta/muse-spark-open",
        "ids": [
          "meta/muse-spark-open"
        ],
        "usd_per_m_in": 1.25,
        "usd_per_m_out": 4.25,
        "usd_per_m_cached_in": 0.15,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "dawncr0w/MiMo-V2.5-oQ4-MLX",
      "family": "mimo-v2-5",
      "member": "base~dawncr0w-oq4",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "XiaomiMiMo/MiMo-V2.5",
      "hub_repo": "XiaomiMiMo/MiMo-V2.5",
      "publisher": "dawncr0w",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 171753,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.5",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.5",
        "ids": [
          "xiaomi/mimo-v2.5"
        ],
        "usd_per_m_in": 0.4,
        "usd_per_m_out": 2,
        "usd_per_m_cached_in": 0.08,
        "context_length": 1000000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "dealignai/DeepSeek-V4.1-Flash-1017-UNCENSORED-FP8",
      "family": "deepseek-v4-1",
      "member": "flash-1017~dealignai-uncensored",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "hub_repo": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "publisher": "dealignai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 486656,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4.1-Flash-1017",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4.1-flash-1017",
        "ids": [
          "deepseek/deepseek-v4.1-flash-1017",
          "deepseek/deepseek-v4.1-flash-1017:batch"
        ],
        "usd_per_m_in": 0.02,
        "usd_per_m_out": 0.108,
        "usd_per_m_cached_in": 0.007,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.336,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 488.0475,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "dealignai/DeepSeek-V4.1-Flash-UNCENSORED-FP8",
      "family": "deepseek-v4-1",
      "member": "flash~dealignai-uncensored",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4.1-Flash",
      "hub_repo": "deepseek-ai/DeepSeek-V4.1-Flash",
      "publisher": "dealignai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 486656,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4.1-Flash",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4.1-flash",
        "ids": [
          "deepseek/deepseek-v4.1-flash",
          "deepseek/deepseek-v4.1-flash:batch"
        ],
        "usd_per_m_in": 0.0507,
        "usd_per_m_out": 0.2028,
        "usd_per_m_cached_in": 0.001014,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.336,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 488.0475,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "dealignai/GLM-5.3-CYBERSECURITY-FP8",
      "family": "glm-5-3",
      "member": "base~jangq-ai",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "JANGQ-AI/GLM-5.3-FP8",
      "hub_repo": "JANGQ-AI/GLM-5.3-FP8",
      "publisher": "dealignai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 720626,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "glm-5.3",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.3",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.3",
        "ids": [
          "z-ai/glm-5.3",
          "z-ai/glm-5.3:batch"
        ],
        "usd_per_m_in": 0.084,
        "usd_per_m_out": 3.6,
        "usd_per_m_cached_in": 0.079,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 997.1067,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "dealignai/MiMo-V2.6-Flash-RL-UNCENSORED",
      "family": "mimo-v2-6",
      "member": "flash-rl~dealignai-uncensored",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "hub_repo": "XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "publisher": "dealignai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 169507,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.6-Flash-RL",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.6-flash",
        "ids": [
          "xiaomi/mimo-v2.6-flash"
        ],
        "usd_per_m_in": 0.14,
        "usd_per_m_out": 0.28,
        "usd_per_m_cached_in": 0.0028,
        "context_length": 1050000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 83.7157,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "deepseek-ai/DeepSeek-V3.2",
      "family": "deepseek-v3-2",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V3.2",
      "hub_repo": "deepseek-ai/DeepSeek-V3.2",
      "publisher": "deepseek-ai",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 657542,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.2",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v3.2",
        "ids": [
          "deepseek/deepseek-v3.2"
        ],
        "usd_per_m_in": 0.26,
        "usd_per_m_out": 0.38,
        "usd_per_m_cached_in": 0.13,
        "context_length": 163840,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 839.7881,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "deepseek-ai/DeepSeek-V3.2-Exp",
      "family": "deepseek-v3-2",
      "member": "exp",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V3.2-Exp",
      "hub_repo": "deepseek-ai/DeepSeek-V3.2-Exp",
      "publisher": "deepseek-ai",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 657542,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.2-Exp",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v3.2-exp",
        "ids": [
          "deepseek/deepseek-v3.2-exp"
        ],
        "usd_per_m_in": 0.27,
        "usd_per_m_out": 0.41,
        "usd_per_m_cached_in": null,
        "context_length": 163840,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 839.7881,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "deepseek-ai/DeepSeek-V4-Flash-0731",
      "family": "deepseek-v4",
      "member": "flash-0731",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4-Flash-0731",
      "hub_repo": "deepseek-ai/DeepSeek-V4-Flash-0731",
      "publisher": "deepseek-ai",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 159155,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4-Flash-0731",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4-flash-0731",
        "ids": [
          "deepseek/deepseek-v4-flash-0731",
          "deepseek/deepseek-v4-flash-0731:batch"
        ],
        "usd_per_m_in": 0.0434,
        "usd_per_m_out": 0.3,
        "usd_per_m_cached_in": 0.035,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.144,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 76.0405,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "deepseek-ai/DeepSeek-V4-Flash-Vision-Exp",
      "family": "deepseek-v4",
      "member": "flash-vision-exp",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4-Flash-Vision-Exp",
      "hub_repo": "deepseek-ai/DeepSeek-V4-Flash-Vision-Exp",
      "publisher": "deepseek-ai",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 160045,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4-Flash-Vision-Exp",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4-flash-vision-exp",
        "ids": [
          "deepseek/deepseek-v4-flash-vision-exp"
        ],
        "usd_per_m_in": 0.44,
        "usd_per_m_out": 1.32,
        "usd_per_m_cached_in": 0.014,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 76.7532,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "deepseek-ai/DeepSeek-V4-Pro",
      "family": "deepseek-v4",
      "member": "pro",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4-Pro",
      "hub_repo": "deepseek-ai/DeepSeek-V4-Pro",
      "publisher": "deepseek-ai",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 824662,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4-Pro",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4-pro",
        "ids": [
          "deepseek/deepseek-v4-pro"
        ],
        "usd_per_m_in": 0.9,
        "usd_per_m_out": 1.5,
        "usd_per_m_cached_in": 0.25,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1283.0155,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "deepseek-ai/DeepSeek-V4-Pro-0813",
      "family": "deepseek-v4",
      "member": "pro-0813",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4-Pro-0813",
      "hub_repo": "deepseek-ai/DeepSeek-V4-Pro-0813",
      "publisher": "deepseek-ai",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 851387,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4-Pro-0813",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4-pro-0813",
        "ids": [
          "deepseek/deepseek-v4-pro-0813"
        ],
        "usd_per_m_in": 0.111,
        "usd_per_m_out": 4.44,
        "usd_per_m_cached_in": 0.097,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.59,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1363.37,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "deepseek-ai/DeepSeek-V4.1-Flash",
      "family": "deepseek-v4-1",
      "member": "flash",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4.1-Flash",
      "hub_repo": "deepseek-ai/DeepSeek-V4.1-Flash",
      "publisher": "deepseek-ai",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 486656,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4.1-Flash",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4.1-flash",
        "ids": [
          "deepseek/deepseek-v4.1-flash",
          "deepseek/deepseek-v4.1-flash:batch"
        ],
        "usd_per_m_in": 0.0507,
        "usd_per_m_out": 0.2028,
        "usd_per_m_cached_in": 0.001014,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.336,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 488.0475,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "family": "deepseek-v4-1",
      "member": "flash-1017",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "hub_repo": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "publisher": "deepseek-ai",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 486656,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4.1-Flash-1017",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4.1-flash-1017",
        "ids": [
          "deepseek/deepseek-v4.1-flash-1017",
          "deepseek/deepseek-v4.1-flash-1017:batch"
        ],
        "usd_per_m_in": 0.02,
        "usd_per_m_out": 0.108,
        "usd_per_m_cached_in": 0.007,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.336,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 488.0475,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "depth-anything/Depth-Anything-V2-Base-hf",
      "family": "depth-anything",
      "member": "v2-base",
      "quant": "FP32",
      "engine": "vision-task",
      "modality": "vision",
      "base": null,
      "hub_repo": "depth-anything/Depth-Anything-V2-Base-hf",
      "publisher": "depth-anything",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 371,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/depth-anything/Depth-Anything-V2-Base-hf",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W36",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T06:23:54.214Z",
          "last_at": "2026-11-14T21:38:52.035Z",
          "caps": [
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T09:58:39.660Z",
          "last_at": "2026-10-22T08:18:20.813Z",
          "caps": [
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:17:50.190Z",
          "last_at": "2026-11-07T02:43:11.525Z",
          "caps": [
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/amd/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-28T13:22:06.050Z",
          "last_at": "2026-10-28T13:22:06.050Z",
          "caps": [
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-27T11:18:15.444Z",
          "last_at": "2026-11-14T13:16:15.933Z",
          "caps": [
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W36"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-03T03:21:07.517Z",
          "last_at": "2026-09-03T03:21:07.517Z",
          "caps": [
            "vision.spatial"
          ],
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-08T19:04:06.193Z",
          "last_at": "2026-11-08T19:04:06.193Z",
          "caps": [
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "depth-anything/Depth-Anything-V2-Small-hf",
      "family": "depth-anything",
      "member": "v2-small",
      "quant": "FP32",
      "engine": "vision-task",
      "modality": "vision",
      "base": null,
      "hub_repo": "depth-anything/Depth-Anything-V2-Small-hf",
      "publisher": "depth-anything",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 94,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/depth-anything/Depth-Anything-V2-Small-hf",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W36",
        "2026-W37",
        "2026-W38",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T06:19:47.255Z",
          "last_at": "2026-11-14T22:40:30.949Z",
          "caps": [
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T06:57:59.068Z",
          "last_at": "2026-10-21T06:57:59.068Z",
          "caps": [
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:18:12.179Z",
          "last_at": "2026-11-07T02:43:30.231Z",
          "caps": [
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/amd/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-28T13:24:41.440Z",
          "last_at": "2026-10-28T13:24:41.440Z",
          "caps": [
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37",
            "2026-W43",
            "2026-W44",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-13T17:49:44.761Z",
          "last_at": "2026-11-14T13:18:34.441Z",
          "caps": [
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W36"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-05T20:29:20.301Z",
          "last_at": "2026-09-05T20:29:20.301Z",
          "caps": [
            "vision.spatial"
          ],
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38",
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-14T06:46:11.768Z",
          "last_at": "2026-11-08T19:07:48.995Z",
          "caps": [
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "dshive/MiMo-V2.5-Pro",
      "family": "mimo-v2-5",
      "member": "pro~dshive",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": null,
      "hub_repo": "dshive/MiMo-V2.5-Pro",
      "publisher": "dshive",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 985517,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/dshive/MiMo-V2.5-Pro",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.5-pro",
        "ids": [
          "xiaomi/mimo-v2.5-pro"
        ],
        "usd_per_m_in": 0.435,
        "usd_per_m_out": 0.87,
        "usd_per_m_cached_in": 0.0036,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1791.2619,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "exalandru/Mistral-Small-4-119B-2603-Mixed-3bit-MLX",
      "family": "mistral-small-4",
      "member": "119b-2603~exalandru-mixed",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "mistralai/Mistral-Small-4-119B-2603",
      "hub_repo": "mistralai/Mistral-Small-4-119B-2603",
      "publisher": "exalandru",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 46074,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Small-4-119B-2603",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-small-2603",
        "ids": [
          "mistralai/mistral-small-2603"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": 0.015,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "facebook/mms-tts-asm",
      "family": "mms-tts",
      "member": "as",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "facebook/mms-tts-asm",
      "hub_repo": "facebook/mms-tts-asm",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 139,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/mms-tts-asm",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T21:36:52.463Z",
          "last_at": "2026-10-27T12:08:44.280Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/mms-tts-ben",
      "family": "mms-tts",
      "member": "bn",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "facebook/mms-tts-ben",
      "hub_repo": "facebook/mms-tts-ben",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 139,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/mms-tts-ben",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T21:24:21.941Z",
          "last_at": "2026-10-27T12:02:45.086Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/mms-tts-guj",
      "family": "mms-tts",
      "member": "gu",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "facebook/mms-tts-guj",
      "hub_repo": "facebook/mms-tts-guj",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 139,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/mms-tts-guj",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T21:31:00.974Z",
          "last_at": "2026-10-27T12:05:19.858Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/mms-tts-hin",
      "family": "mms-tts",
      "member": "hi",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "facebook/mms-tts-hin",
      "hub_repo": "facebook/mms-tts-hin",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 139,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/mms-tts-hin",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T21:23:06.949Z",
          "last_at": "2026-10-27T12:02:03.899Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-06T09:29:06.742Z",
          "last_at": "2026-11-06T09:29:06.742Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/amd/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-28T12:02:55.721Z",
          "last_at": "2026-10-28T12:02:55.721Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-07T20:51:54.282Z",
          "last_at": "2026-11-14T12:28:17.393Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-08T15:09:20.684Z",
          "last_at": "2026-11-08T15:09:20.684Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/mms-tts-kan",
      "family": "mms-tts",
      "member": "kn",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "facebook/mms-tts-kan",
      "hub_repo": "facebook/mms-tts-kan",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 139,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/mms-tts-kan",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T21:32:05.521Z",
          "last_at": "2026-10-27T12:05:55.617Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/mms-tts-mal",
      "family": "mms-tts",
      "member": "ml",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "facebook/mms-tts-mal",
      "hub_repo": "facebook/mms-tts-mal",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 139,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/mms-tts-mal",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T21:33:16.708Z",
          "last_at": "2026-10-27T12:06:35.047Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/mms-tts-mar",
      "family": "mms-tts",
      "member": "mr",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "facebook/mms-tts-mar",
      "hub_repo": "facebook/mms-tts-mar",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 139,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/mms-tts-mar",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T21:29:55.555Z",
          "last_at": "2026-10-27T12:04:43.537Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/mms-tts-ory",
      "family": "mms-tts",
      "member": "or",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "facebook/mms-tts-ory",
      "hub_repo": "facebook/mms-tts-ory",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 139,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/mms-tts-ory",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T21:35:43.973Z",
          "last_at": "2026-10-27T12:08:07.736Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/mms-tts-pan",
      "family": "mms-tts",
      "member": "pa",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "facebook/mms-tts-pan",
      "hub_repo": "facebook/mms-tts-pan",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 139,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/mms-tts-pan",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T21:34:29.006Z",
          "last_at": "2026-10-27T12:07:16.475Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/mms-tts-tam",
      "family": "mms-tts",
      "member": "ta",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "facebook/mms-tts-tam",
      "hub_repo": "facebook/mms-tts-tam",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 139,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/mms-tts-tam",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T21:27:22.973Z",
          "last_at": "2026-10-27T12:03:59.964Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/mms-tts-tel",
      "family": "mms-tts",
      "member": "te",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "facebook/mms-tts-tel",
      "hub_repo": "facebook/mms-tts-tel",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 139,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/mms-tts-tel",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T21:25:32.643Z",
          "last_at": "2026-10-27T12:03:21.754Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/mms-tts-urd-script_arabic",
      "family": "mms-tts",
      "member": "ur",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "facebook/mms-tts-urd-script_arabic",
      "hub_repo": "facebook/mms-tts-urd-script_arabic",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 139,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/mms-tts-urd-script_arabic",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T21:38:10.023Z",
          "last_at": "2026-10-27T12:09:39.060Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/musicgen-medium",
      "family": "musicgen",
      "member": "medium",
      "quant": "FP32",
      "engine": "audiogen",
      "modality": "audio",
      "base": null,
      "hub_repo": "facebook/musicgen-medium",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 1,
      "rung": 1,
      "disk_mib": 11402,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/musicgen-medium",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T18:28:25.792Z",
          "last_at": "2026-10-22T09:35:49.278Z",
          "caps": [
            "audio.music"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:21:35.938Z",
          "last_at": "2026-11-07T02:58:25.849Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/musicgen-small",
      "family": "musicgen",
      "member": "small",
      "quant": "FP32",
      "engine": "audiogen",
      "modality": "audio",
      "base": null,
      "hub_repo": "facebook/musicgen-small",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 2,
      "rung": 1,
      "disk_mib": 2254,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/musicgen-small",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W36",
        "2026-W37",
        "2026-W38",
        "2026-W39",
        "2026-W40",
        "2026-W41",
        "2026-W43",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38",
            "2026-W40",
            "2026-W41"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-16T06:43:08.948Z",
          "last_at": "2026-10-10T21:33:47.641Z",
          "caps": [
            "audio.music"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "darwin/arm64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-07T04:15:48.177Z",
          "last_at": "2026-09-07T04:15:48.177Z",
          "caps": [
            "audio.music"
          ],
          "verdict": "only the network runs it"
        },
        "darwin/arm64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W36"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-06T21:46:31.373Z",
          "last_at": "2026-09-06T21:46:31.373Z",
          "caps": [
            "audio.music"
          ],
          "verdict": "only the network runs it"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-05T11:16:53.075Z",
          "last_at": "2026-11-14T22:39:24.329Z",
          "caps": [
            "audio.music"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T06:30:29.971Z",
          "last_at": "2026-10-22T07:59:28.315Z",
          "caps": [
            "audio.music"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:20:31.842Z",
          "last_at": "2026-11-07T02:55:08.605Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/amd/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-17T08:58:26.913Z",
          "last_at": "2026-09-17T08:58:26.913Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W37",
            "2026-W39",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-13T15:47:28.143Z",
          "last_at": "2026-11-13T23:07:32.401Z",
          "caps": [
            "audio.music"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W36",
            "2026-W37"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-01T10:39:19.915Z",
          "last_at": "2026-09-08T14:25:44.562Z",
          "caps": [
            "audio.music"
          ],
          "verdict": "only the network runs it"
        },
        "windows/amd64/none/8g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-09T11:55:29.794Z",
          "last_at": "2026-09-09T12:12:59.385Z",
          "caps": [
            "audio.music"
          ],
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38",
            "2026-W40"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T14:56:12.676Z",
          "last_at": "2026-10-02T17:12:56.374Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-16T12:17:47.739Z",
          "last_at": "2026-09-16T12:17:47.739Z",
          "caps": [],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/nllb-200-distilled-1.3B",
      "family": "nllb",
      "member": "1.3b-distilled",
      "quant": "FP32",
      "engine": "translate-hf",
      "modality": "translation",
      "base": "facebook/nllb-200-distilled-1.3B",
      "hub_repo": "facebook/nllb-200-distilled-1.3B",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 5228,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/nllb-200-distilled-1.3B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W44",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43",
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T23:26:24.053Z",
          "last_at": "2026-10-27T12:48:56.710Z",
          "caps": [
            "translate.as",
            "translate.bn",
            "translate.gu",
            "translate.hi",
            "translate.kn",
            "translate.ml",
            "translate.mr",
            "translate.or",
            "translate.pa",
            "translate.ta",
            "translate.te",
            "translate.text",
            "translate.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T21:59:06.846Z",
          "last_at": "2026-11-07T02:49:47.929Z",
          "caps": [
            "translate.as",
            "translate.bn",
            "translate.gu",
            "translate.hi",
            "translate.kn",
            "translate.ml",
            "translate.mr",
            "translate.or",
            "translate.pa",
            "translate.ta",
            "translate.te",
            "translate.text",
            "translate.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-08T19:42:05.140Z",
          "last_at": "2026-11-08T19:42:05.140Z",
          "caps": [
            "translate.as",
            "translate.gu",
            "translate.hi",
            "translate.text",
            "translate.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/nllb-200-distilled-600M",
      "family": "nllb",
      "member": "600m-distilled",
      "quant": "FP32",
      "engine": "translate-hf",
      "modality": "translation",
      "base": "facebook/nllb-200-distilled-600M",
      "hub_repo": "facebook/nllb-200-distilled-600M",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 2346,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/nllb-200-distilled-600M",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:34:36.372Z",
          "last_at": "2026-11-14T21:50:20.448Z",
          "caps": [
            "translate.as",
            "translate.bn",
            "translate.gu",
            "translate.hi",
            "translate.kn",
            "translate.ml",
            "translate.mr",
            "translate.or",
            "translate.pa",
            "translate.ta",
            "translate.te",
            "translate.text",
            "translate.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T21:56:57.411Z",
          "last_at": "2026-11-07T02:46:04.134Z",
          "caps": [
            "translate.as",
            "translate.bn",
            "translate.gu",
            "translate.hi",
            "translate.kn",
            "translate.ml",
            "translate.mr",
            "translate.or",
            "translate.pa",
            "translate.ta",
            "translate.te",
            "translate.text",
            "translate.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-28T05:57:27.982Z",
          "last_at": "2026-11-07T20:26:37.160Z",
          "caps": [
            "translate.as",
            "translate.bn",
            "translate.gu",
            "translate.hi",
            "translate.kn",
            "translate.ml",
            "translate.mr",
            "translate.or",
            "translate.pa",
            "translate.ta",
            "translate.te",
            "translate.text",
            "translate.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-10T09:40:00.877Z",
          "last_at": "2026-11-10T09:40:00.877Z",
          "caps": [
            "translate.as",
            "translate.bn",
            "translate.gu",
            "translate.hi",
            "translate.kn",
            "translate.ml",
            "translate.mr",
            "translate.or",
            "translate.pa",
            "translate.ta",
            "translate.te",
            "translate.text",
            "translate.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/sam-vit-base",
      "family": "sam",
      "member": "vit-base",
      "quant": "FP32",
      "engine": "vision-task",
      "modality": "vision",
      "base": null,
      "hub_repo": "facebook/sam-vit-base",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 715,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/sam-vit-base",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W36",
        "2026-W37",
        "2026-W38",
        "2026-W40",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W40"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-01T08:42:58.547Z",
          "last_at": "2026-10-01T08:42:58.547Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T06:23:26.300Z",
          "last_at": "2026-11-14T21:29:30.817Z",
          "caps": [
            "vision.count"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T07:03:41.914Z",
          "last_at": "2026-10-22T08:16:04.272Z",
          "caps": [
            "vision.count"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T21:23:44.842Z",
          "last_at": "2026-11-06T17:10:46.263Z",
          "caps": [
            "vision.count"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/amd/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-28T12:16:19.470Z",
          "last_at": "2026-10-28T12:16:19.470Z",
          "caps": [
            "vision.count"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-13T18:09:11.329Z",
          "last_at": "2026-11-15T13:30:31.227Z",
          "caps": [
            "vision.count"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W36"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-05T11:25:26.123Z",
          "last_at": "2026-09-05T11:25:26.123Z",
          "caps": [
            "vision.count"
          ],
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-14T05:53:34.018Z",
          "last_at": "2026-09-14T05:53:34.018Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "facebook/seamless-m4t-v2-large",
      "family": "seamless-m4t",
      "member": "v2-large",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "facebook/seamless-m4t-v2-large",
      "hub_repo": "facebook/seamless-m4t-v2-large",
      "publisher": "facebook",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 8840,
      "active_share": 1,
      "licence": {
        "id": "cc-by-nc-4.0",
        "name": null,
        "class": "non-commercial",
        "serve_scopes": [
          "own"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/facebook/seamless-m4t-v2-large",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:06:32.927Z",
          "last_at": "2026-11-06T18:06:00.315Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ne",
            "lang.or",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "lang.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-09T11:55:59.772Z",
          "last_at": "2026-11-14T13:02:09.694Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ne",
            "lang.or",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "lang.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-08T18:03:31.419Z",
          "last_at": "2026-11-08T18:44:18.195Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ne",
            "lang.or",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "lang.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "fenyo/4xH100-Mistral-Medium-3.5-128B-MonEspaceSante",
      "family": "mistral-medium-3-5",
      "member": "128b~fenyo-4xh100-monespacesante",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "mistralai/Mistral-Medium-3.5-128B",
      "hub_repo": "mistralai/Mistral-Medium-3.5-128B",
      "publisher": "fenyo",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 126058,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Medium-3.5-128B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-medium-3-5",
        "ids": [
          "mistralai/mistral-medium-3-5",
          "mistralai/mistral-medium-3-5:batch"
        ],
        "usd_per_m_in": 1.5,
        "usd_per_m_out": 7.5,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": 3.75,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 53.5285,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "finding1/DeepSeek-V3.1-Terminus-MLX-mixed_4_6",
      "family": "deepseek-v3-1",
      "member": "terminus~finding1-mixed-4-6",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V3.1-Terminus",
      "hub_repo": "deepseek-ai/DeepSeek-V3.1-Terminus",
      "publisher": "finding1",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 384716,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.1-Terminus",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v3.1-terminus",
        "ids": [
          "deepseek/deepseek-v3.1-terminus"
        ],
        "usd_per_m_in": 0.27,
        "usd_per_m_out": 1,
        "usd_per_m_cached_in": null,
        "context_length": 163840,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "finding1/DeepSeek-V3.1-Terminus-MLX-mixed_4_6-v0.28.2",
      "family": "deepseek-v3-1",
      "member": "terminus~finding1-mixed-4-6-28-2",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V3.1-Terminus",
      "hub_repo": "deepseek-ai/DeepSeek-V3.1-Terminus",
      "publisher": "finding1",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 384865,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.1-Terminus",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v3.1-terminus",
        "ids": [
          "deepseek/deepseek-v3.1-terminus"
        ],
        "usd_per_m_in": 0.27,
        "usd_per_m_out": 1,
        "usd_per_m_cached_in": null,
        "context_length": 163840,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "fno2010/MiniMax-M2.7-TQ3",
      "family": "minimax-m2-7",
      "member": "base~fno2010-tq3",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.7",
      "hub_repo": "MiniMaxAI/MiniMax-M2.7",
      "publisher": "fno2010",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 90516,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "other",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.7",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.7",
        "ids": [
          "minimax/minimax-m2.7"
        ],
        "usd_per_m_in": 0.21,
        "usd_per_m_out": 0.84,
        "usd_per_m_cached_in": 0.042,
        "context_length": 196608,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 33.2477,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "gemma4:e2b",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E2B-it",
      "hub_repo": "google/gemma-4-E2B-it",
      "publisher": "google",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 7162,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E2B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W31",
        "2026-W34",
        "2026-W38",
        "2026-W43",
        "2026-W44",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.3229
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 28.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.6778,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 7.15,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 4.689,
          "first_at": "2026-10-27T02:59:42.265Z",
          "last_at": "2026-10-27T02:59:42.265Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 5.0692,
          "usd_per_m_energy": 0.1303,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 190.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.5343,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43",
            "2026-W45"
          ],
          "tokens_per_sec_best": 112.24,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.114,
          "first_at": "2026-10-20T09:24:28.225Z",
          "last_at": "2026-11-03T02:23:43.394Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.ar",
            "lang.as",
            "lang.bn",
            "lang.de",
            "lang.en",
            "lang.es",
            "lang.fa",
            "lang.fr",
            "lang.gu",
            "lang.he",
            "lang.hi",
            "lang.id",
            "lang.it",
            "lang.ja",
            "lang.kn",
            "lang.ko",
            "lang.ml",
            "lang.mr",
            "lang.ms",
            "lang.ne",
            "lang.nl",
            "lang.or",
            "lang.pa",
            "lang.pl",
            "lang.pt",
            "lang.ru",
            "lang.si",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "lang.th",
            "lang.tr",
            "lang.uk",
            "lang.ur",
            "lang.vi",
            "lang.zh",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.3229,
          "usd_per_m_energy": 0.0309,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 149.89,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 0.732,
          "first_at": "2026-11-07T00:35:39.324Z",
          "last_at": "2026-11-07T00:35:39.324Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.7829,
          "usd_per_m_energy": 0.0203,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "unknown": {
          "proved": true,
          "grams": 3,
          "seasons": [
            "2026-W31",
            "2026-W34"
          ],
          "tokens_per_sec_best": 43.34,
          "effective_ctx_best": 8192,
          "joules_per_token_best": null,
          "first_at": "2026-07-28T15:50:32.066Z",
          "last_at": "2026-08-21T07:03:15.484Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.09,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 70.4593,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 41.0699,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 170.45,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T14:48:40.215Z",
          "last_at": "2026-09-15T14:48:40.215Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.5376,
          "usd_per_m_energy": 0.0367,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "gemma4:e2b-it-qat",
      "family": "gemma-4",
      "member": "e2b",
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": null,
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W29"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W29"
          ],
          "tokens_per_sec_best": 8.93,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-07-14T21:10:37.731Z",
          "last_at": "2026-07-14T21:10:37.731Z",
          "caps": [],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "gemma4:e4b",
      "family": "gemma-4",
      "member": "e4b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": null,
      "publisher": "google",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 11000,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.4661
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 18.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.2809,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.1342,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 123.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.9177,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 84.47,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 2.303,
          "first_at": "2026-10-23T06:30:14.408Z",
          "last_at": "2026-10-24T16:31:18.583Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.4661,
          "usd_per_m_energy": 0.064,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.9321,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 112.5691,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.22,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 65.5115,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 62.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.2142,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "gemma4:e4b-it-qat",
      "family": "gemma-4",
      "member": "e4b",
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": null,
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W29"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W29"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-07-14T21:08:25.541Z",
          "last_at": "2026-07-14T21:08:25.541Z",
          "caps": [],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "google/gemma-4-E2B-it-qat-q4_0-gguf",
      "family": "gemma-4",
      "member": "e2b",
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "google/gemma-4-E2B-it-qat-q4_0-gguf",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W29"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W29"
          ],
          "tokens_per_sec_best": 1.99,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-07-14T10:21:08.678Z",
          "last_at": "2026-07-14T10:21:08.678Z",
          "caps": [
            "structured_output"
          ],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "google/madlad400-3b-mt",
      "family": "madlad",
      "member": "3b",
      "quant": "FP32",
      "engine": "translate-hf",
      "modality": "translation",
      "base": "google/madlad400-3b-mt",
      "hub_repo": "google/madlad400-3b-mt",
      "publisher": "google",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 11216,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/madlad400-3b-mt",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W44",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43",
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T21:25:15.957Z",
          "last_at": "2026-10-27T13:24:56.870Z",
          "caps": [
            "translate.as",
            "translate.bn",
            "translate.gu",
            "translate.hi",
            "translate.kn",
            "translate.ml",
            "translate.mr",
            "translate.or",
            "translate.pa",
            "translate.ta",
            "translate.text"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:11:22.484Z",
          "last_at": "2026-11-06T18:10:50.632Z",
          "caps": [
            "translate.as",
            "translate.bn",
            "translate.gu",
            "translate.hi",
            "translate.kn",
            "translate.ml",
            "translate.mr",
            "translate.or",
            "translate.pa",
            "translate.ta",
            "translate.text"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "helixdouble/GLM-5.1-Abliterated",
      "family": "glm-5-1",
      "member": "base~helixdouble-abliterated",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-5.1-FP8",
      "hub_repo": "zai-org/GLM-5.1-FP8",
      "publisher": "helixdouble",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 721147,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.1-FP8",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.1",
        "ids": [
          "z-ai/glm-5.1"
        ],
        "usd_per_m_in": 1.4,
        "usd_per_m_out": 4.4,
        "usd_per_m_cached_in": 0.26,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 997.7124,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "hf.co/0bserverx/Muse-Glimmer-30B-Heretic-Uncensored-GGUF:Q4_K_M",
      "family": "muse-glimmer",
      "member": "30b~0bserverx-heretic-uncensored",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-models/Muse-Glimmer-30B",
      "hub_repo": "meta-models/Muse-Glimmer-30B",
      "publisher": "0bserverx",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 16150,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/meta-models/Muse-Glimmer-30B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta/muse-glimmer-30b",
        "ids": [
          "meta/muse-glimmer-30b",
          "meta/muse-glimmer-30b:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.04,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.96,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.6128,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.8667,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 84.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.8011,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 29.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4913,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.5296,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 175.2398,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 101.3435,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 42.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.2854,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Anbeeld/Kimi-K2.6-DSpark-GGUF:Q4_K_M",
      "family": "kimi-k2-6",
      "member": "base~anbeeld-dspark",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.6",
      "hub_repo": "moonshotai/Kimi-K2.6",
      "publisher": "Anbeeld",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 1787,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.6",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.6",
        "ids": [
          "moonshotai/kimi-k2.6"
        ],
        "usd_per_m_in": 0.65,
        "usd_per_m_out": 3.41,
        "usd_per_m_cached_in": 0.15,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 114.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6302,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 26.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2022,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 762.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6266,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 263.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1207,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 106.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0563,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 16.6116,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.6569,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 387.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6653,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Anbeeld/Kimi-K2.7-Code-DSpark-GGUF:Q4_K_M",
      "family": "kimi-k2-7",
      "member": "code~anbeeld-dspark",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.7-Code",
      "hub_repo": "moonshotai/Kimi-K2.7-Code",
      "publisher": "Anbeeld",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 1787,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.7-Code",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.7-code",
        "ids": [
          "moonshotai/kimi-k2.7-code"
        ],
        "usd_per_m_in": 0.7125,
        "usd_per_m_out": 3.5,
        "usd_per_m_cached_in": 0.16,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 114.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6302,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 26.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2022,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 762.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6266,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 263.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1207,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 106.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0563,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 16.6116,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.6569,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 387.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6653,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Anbeeld/MiniMax-M3-DSpark-GGUF:Q4_K_M",
      "family": "minimax-m3",
      "member": "base~anbeeld-dspark",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M3",
      "hub_repo": "MiniMaxAI/MiniMax-M3",
      "publisher": "Anbeeld",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 1156,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "minimax-community",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M3",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m3",
        "ids": [
          "minimax/minimax-m3",
          "minimax/minimax-m3:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.06,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.88,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 177.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4048,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 40.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7652,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1179.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4049,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 408.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0768,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 165.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6802,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.6696,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.59,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.2047,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 598.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4295,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/AngelSlim/Hy3-GGUF:Q4_K_M",
      "family": "hy-3",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "tencent/Hy3",
      "hub_repo": "tencent/Hy3",
      "publisher": "AngelSlim",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 173881,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy3",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy3",
        "ids": [
          "tencent/hy3"
        ],
        "usd_per_m_in": 0.0825,
        "usd_per_m_out": 0.33,
        "usd_per_m_cached_in": 0.020625,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 180.0652,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 636.7459,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 78.8307,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 63.4396,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 230.0295,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5311.9236,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2655.9618,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 99.9384,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/AtomicChat/LFM2.5-24B-A2B-Uncensored-GGUF:Q4_K_M",
      "family": "lfm-2-5",
      "member": "24b-a2b~atomicchat-uncensored",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "LiquidAI/LFM2.5-24B-A2B",
      "hub_repo": "LiquidAI/LFM2.5-24B-A2B",
      "publisher": "AtomicChat",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 14648,
      "active_share": 0.0833,
      "licence": {
        "id": "other",
        "name": "lfm1.0",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/LiquidAI/LFM2.5-24B-A2B",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "liquid/lfm-2.5-24b-a2b",
        "ids": [
          "liquid/lfm-2.5-24b-a2b"
        ],
        "usd_per_m_in": 0.05,
        "usd_per_m_out": 0.7,
        "usd_per_m_cached_in": 0.025,
        "context_length": 128000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 168.25,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4927,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 38.81,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0909,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1116.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4374,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 386.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1096,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 156.59,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7884,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.9926,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.97,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.5563,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 567.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4729,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/DavidAU/OpenAi-GPT-oss-20b-HERETIC-uncensored-NEO-Imatrix-gguf:Q8_0",
      "family": "gpt-oss",
      "member": "20b~p-e-w-heretic",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "p-e-w/gpt-oss-20b-heretic",
      "hub_repo": "p-e-w/gpt-oss-20b-heretic",
      "publisher": "DavidAU",
      "derivative": true,
      "cohort": 3,
      "rung": 1,
      "disk_mib": 21058,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/p-e-w/gpt-oss-20b-heretic",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 14.1529
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.0368,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 3.36,
          "effective_ctx_best": null,
          "joules_per_token_best": 26.173,
          "first_at": "2026-10-28T08:06:25.364Z",
          "last_at": "2026-10-28T08:06:25.364Z",
          "caps": [],
          "usd_per_m_out": 14.1529,
          "usd_per_m_energy": 0.727,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 64.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.6265,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.4,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.1229,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.1701,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.37,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 238.1332,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 137.6707,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 32.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.3172,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/DavidAU/OpenAi-GPT-oss-20b-abliterated-uncensored-NEO-Imatrix-gguf:Q8_0",
      "family": "gpt-oss",
      "member": "20b~huihui-ai-abliterated",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "huihui-ai/Huihui-gpt-oss-20b-BF16-abliterated",
      "hub_repo": "huihui-ai/Huihui-gpt-oss-20b-BF16-abliterated",
      "publisher": "DavidAU",
      "derivative": true,
      "cohort": 3,
      "rung": 1,
      "disk_mib": 21058,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/huihui-ai/Huihui-gpt-oss-20b-BF16-abliterated",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 14.4981
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.0368,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 3.28,
          "effective_ctx_best": null,
          "joules_per_token_best": 9.683,
          "first_at": "2026-10-28T07:30:01.792Z",
          "last_at": "2026-10-28T07:30:01.792Z",
          "caps": [],
          "usd_per_m_out": 14.4981,
          "usd_per_m_energy": 0.269,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 64.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.6265,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.4,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.1229,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.1701,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.37,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 238.1332,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 137.6707,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 32.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.3172,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/EspenJD/Step-3.7-Flash-Q4_K_M-MTP-GGUF:Q4_K_M",
      "family": "step-3-7",
      "member": "flash~espenjd-k-m-mtp",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "stepfun-ai/Step-3.7-Flash",
      "hub_repo": "stepfun-ai/Step-3.7-Flash",
      "publisher": "EspenJD",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 113089,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "stepfun/step-3.7-flash",
        "ids": [
          "stepfun/step-3.7-flash"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 1.15,
        "usd_per_m_cached_in": 0.04,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 89.5628,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 291.5446,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.05,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 47.1834,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 29.3642,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 120.4496,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2328.6327,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1358.3691,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 56.9088,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/HauhauCS/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-MTP-GGUF:Q6_K_P",
      "family": "qwen-3-8",
      "member": "27b~hauhaucs-uncensored-aggressive-mtp",
      "quant": "Q6_K_P",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "HauhauCS",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 24723,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 28.8779,
        "vs_api": 15.4
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.9617,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.321,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 55.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.007,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 1.75,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 126.583,
          "first_at": "2026-10-28T15:21:23.960Z",
          "last_at": "2026-10-28T15:21:23.960Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call"
          ],
          "usd_per_m_out": 28.8779,
          "usd_per_m_energy": 3.5162,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 17.0307,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.32,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 284.662,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 168.6886,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.8703,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Jackrong/DeepSeek-V4-Pro-Qwen3.5-9B-MTP-GGUF:Q4_K_M",
      "family": "deepseek-v4",
      "member": "9b~unsloth-qwen3-5",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "unsloth/Qwen3.5-9B",
      "hub_repo": "unsloth/Qwen3.5-9B",
      "publisher": "Jackrong",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 5512,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/unsloth/Qwen3.5-9B",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.5-9b",
        "ids": [
          "qwen/qwen3.5-9b"
        ],
        "usd_per_m_in": 0.037,
        "usd_per_m_out": 0.054,
        "usd_per_m_cached_in": 0.033,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 37.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.0252,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.59,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.0631,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 247.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.945,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 85.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4078,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 34.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.3453,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 53.5162,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 31.0526,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 125.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.0763,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Jackrong/Qwen3.5-9B-DeepSeek-V4-Flash-GGUF:Q4_K_M",
      "family": "deepseek-v4",
      "member": "9b~unsloth-qwen3-5",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "unsloth/Qwen3.5-9B",
      "hub_repo": "unsloth/Qwen3.5-9B",
      "publisher": "Jackrong",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 5368,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/unsloth/Qwen3.5-9B",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.5-9b",
        "ids": [
          "qwen/qwen3.5-9b"
        ],
        "usd_per_m_in": 0.037,
        "usd_per_m_out": 0.054,
        "usd_per_m_cached_in": 0.033,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 38.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.9692,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.9439,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 253.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8937,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 87.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3959,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 35.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.2546,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 51.9591,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 30.1363,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 128.95,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.0211,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/JonathanColetti/Qwen3.8-27B-Uncensored-GGUF:Q4_K_M",
      "family": "qwen-3-8",
      "member": "27b~jonathancoletti-uncensored",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "JonathanColetti",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 16031,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 4.4125,
        "vs_api": 2.35
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.81,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.5588,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.95,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.7331,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 85.03,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.7577,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 9.85,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 23.163,
          "first_at": "2026-10-28T12:07:25.615Z",
          "last_at": "2026-10-28T12:07:25.615Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 4.4125,
          "usd_per_m_energy": 0.6434,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.4508,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 171.4659,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 100.0218,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.2366,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/JonathanColetti/Qwen3.8-27B-Uncensored-GGUF:Q5_K_M",
      "family": "qwen-3-8",
      "member": "27b~jonathancoletti-uncensored",
      "quant": "Q5_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "JonathanColetti",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 18630,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 13.5246,
        "vs_api": 7.21
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.8161,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 17.944,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 73.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.7198,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 3.37,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 35.822,
          "first_at": "2026-10-28T13:50:38.058Z",
          "last_at": "2026-10-28T13:50:38.058Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 13.5246,
          "usd_per_m_energy": 0.9951,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.3478,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 205.0794,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 119.6297,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 37.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.3058,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/LiquidAI/LFM2.5-24B-A2B-GGUF:Q4_K_M",
      "family": "lfm-2-5",
      "member": "24b-a2b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "LiquidAI/LFM2.5-24B-A2B",
      "hub_repo": "LiquidAI/LFM2.5-24B-A2B",
      "publisher": "LiquidAI",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 13853,
      "active_share": 0.0833,
      "licence": {
        "id": "other",
        "name": "lfm1.0",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/LiquidAI/LFM2.5-24B-A2B",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "liquid/lfm-2.5-24b-a2b",
        "ids": [
          "liquid/lfm-2.5-24b-a2b"
        ],
        "usd_per_m_in": 0.05,
        "usd_per_m_out": 0.7,
        "usd_per_m_cached_in": 0.025,
        "context_length": 128000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 177.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4623,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 41.03,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0161,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1180.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4131,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 408.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.102,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 165.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7416,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.1846,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.0902,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 599.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4462,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:IQ1_S",
      "family": "llama-3-2",
      "member": "1b",
      "quant": "IQ1_S",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-1B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W34",
        "2026-W35",
        "2026-W37",
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-1b-instruct",
        "ids": [
          "meta-llama/llama-3.2-1b-instruct"
        ],
        "usd_per_m_in": 0.027,
        "usd_per_m_out": 0.201,
        "usd_per_m_cached_in": null,
        "context_length": 60000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/amd64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 3.26,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-08T12:35:58.168Z",
          "last_at": "2026-09-08T12:35:58.168Z",
          "caps": [],
          "verdict": "proved"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 6.6,
          "effective_ctx_best": null,
          "joules_per_token_best": 12.701,
          "first_at": "2026-10-26T05:34:34.037Z",
          "last_at": "2026-10-26T05:34:34.037Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W34",
            "2026-W35"
          ],
          "tokens_per_sec_best": 105.02,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-21T10:27:02.312Z",
          "last_at": "2026-08-26T10:14:14.497Z",
          "caps": [],
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:IQ3_XS",
      "family": "llama-3-2",
      "member": "1b",
      "quant": "IQ3_XS",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-1B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W35",
        "2026-W37"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-1b-instruct",
        "ids": [
          "meta-llama/llama-3.2-1b-instruct"
        ],
        "usd_per_m_in": 0.027,
        "usd_per_m_out": 0.201,
        "usd_per_m_cached_in": null,
        "context_length": 60000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/amd64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 1.48,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-08T13:07:42.920Z",
          "last_at": "2026-09-08T13:07:42.920Z",
          "caps": [
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "verdict": "proved"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 107.53,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-26T10:22:58.777Z",
          "last_at": "2026-08-26T10:22:58.777Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "verdict": "proved"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 17.72,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-09T11:25:34.594Z",
          "last_at": "2026-09-09T11:25:34.594Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.nl",
            "lang.ta",
            "output.schema",
            "structured_output",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
      "family": "llama-3-2",
      "member": "1b",
      "quant": "Q4_K_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/Llama-3.2-1B-Instruct-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-1B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W34",
        "2026-W35"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-1b-instruct",
        "ids": [
          "meta-llama/llama-3.2-1b-instruct"
        ],
        "usd_per_m_in": 0.027,
        "usd_per_m_out": 0.201,
        "usd_per_m_cached_in": null,
        "context_length": 60000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 113.27,
          "effective_ctx_best": 8192,
          "joules_per_token_best": null,
          "first_at": "2026-08-28T09:20:34.027Z",
          "last_at": "2026-08-28T09:20:34.027Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "verdict": "proved"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W34"
          ],
          "tokens_per_sec_best": 8.85,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-23T19:43:25.724Z",
          "last_at": "2026-08-23T19:43:25.724Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/Llama-3.2-3B-Instruct-GGUF:IQ1_S",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "IQ1_S",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/Llama-3.2-3B-Instruct-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W35"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 45.74,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-26T09:46:20.409Z",
          "last_at": "2026-08-26T09:46:20.409Z",
          "caps": [
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "Q4_K_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/Llama-3.2-3B-Instruct-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 78.03,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 6.572,
          "first_at": "2026-11-08T13:25:16.279Z",
          "last_at": "2026-11-08T13:25:16.279Z",
          "caps": [
            "faithful_tool_result",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q2_K",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q2_K",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/Phi-4-mini-instruct-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W38"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 154.35,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T15:47:45.257Z",
          "last_at": "2026-09-15T15:47:45.257Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_L",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q3_K_L",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/Phi-4-mini-instruct-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W35",
        "2026-W38",
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/amd64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 2.2,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-28T05:36:01.320Z",
          "last_at": "2026-08-28T05:36:01.320Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 64.06,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 8.871,
          "first_at": "2026-10-20T12:29:40.657Z",
          "last_at": "2026-10-20T12:29:40.657Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 110.09,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T16:11:00.573Z",
          "last_at": "2026-09-15T16:11:00.573Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q3_K_S",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q3_K_S",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/Phi-4-mini-instruct-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W35",
        "2026-W38"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/amd64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 2.17,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-27T23:41:25.985Z",
          "last_at": "2026-08-27T23:41:25.985Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "verdict": "only the network runs it"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 27.89,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-26T08:31:50.012Z",
          "last_at": "2026-08-26T08:31:50.012Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 134.8,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T15:58:21.029Z",
          "last_at": "2026-09-15T15:58:21.029Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q4_K_M",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q4_K_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/Phi-4-mini-instruct-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W38",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38",
            "2026-W45"
          ],
          "tokens_per_sec_best": 205.49,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 2.094,
          "first_at": "2026-09-15T14:46:09.136Z",
          "last_at": "2026-11-08T15:45:07.861Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.nl",
            "lang.ru",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q4_K_S",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q4_K_S",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/Phi-4-mini-instruct-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W38"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 77.39,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T16:31:44.130Z",
          "last_at": "2026-09-15T16:31:44.130Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/Phi-4-mini-instruct-GGUF:Q8_0",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q8_0",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/Phi-4-mini-instruct-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W38"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 39.21,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T16:54:49.000Z",
          "last_at": "2026-09-15T16:54:49.000Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/Qwen3-0.6B-GGUF:Q3_K_M",
      "family": "qwen-3",
      "member": "0.6b",
      "quant": "Q3_K_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/Qwen3-0.6B-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W34"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W34"
          ],
          "tokens_per_sec_best": 44.56,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-21T12:07:34.663Z",
          "last_at": "2026-08-21T12:07:34.663Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/Qwen3-0.6B-GGUF:Q4_K_M",
      "family": "qwen-3",
      "member": "0.6b",
      "quant": "Q4_K_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/Qwen3-0.6B-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W34"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W34"
          ],
          "tokens_per_sec_best": 37.41,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-21T13:40:21.262Z",
          "last_at": "2026-08-21T13:40:21.262Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/gemma-3-1b-it-GGUF:Q2_K",
      "family": "gemma-3",
      "member": "1b",
      "quant": "Q2_K",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/gemma-3-1b-it-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 12.96,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 2.264,
          "first_at": "2026-10-25T23:09:20.635Z",
          "last_at": "2026-10-25T23:09:20.635Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/gemma-3-1b-it-GGUF:Q3_K_S",
      "family": "gemma-3",
      "member": "1b",
      "quant": "Q3_K_S",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "MaziyarPanahi/gemma-3-1b-it-GGUF",
      "publisher": "maziyarpanahi",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 13.61,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 2.069,
          "first_at": "2026-10-25T22:53:06.021Z",
          "last_at": "2026-10-25T22:53:06.021Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q5_K_M",
      "family": "gpt-oss",
      "member": "20b~arliai-derestricted",
      "quant": "Q5_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "ArliAI/gpt-oss-20b-Derestricted",
      "hub_repo": "ArliAI/gpt-oss-20b-Derestricted",
      "publisher": "MaziyarPanahi",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 16110,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ArliAI/gpt-oss-20b-Derestricted",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 12.6532
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.5947,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 3.44,
          "effective_ctx_best": null,
          "joules_per_token_best": 16.235,
          "first_at": "2026-10-28T06:42:56.397Z",
          "last_at": "2026-10-28T06:42:56.397Z",
          "caps": [],
          "usd_per_m_out": 12.6532,
          "usd_per_m_energy": 0.451,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 84.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.7863,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 29.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4866,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.5091,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 175.172,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 101.3043,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 42.97,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.2686,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/MaziyarPanahi/phi-4-GGUF:Q4_K_M",
      "family": "phi-4",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/phi-4",
      "hub_repo": "microsoft/phi-4",
      "publisher": "MaziyarPanahi",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 8633,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/phi-4",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "microsoft/phi-4",
        "ids": [
          "microsoft/phi-4"
        ],
        "usd_per_m_in": 0.07,
        "usd_per_m_out": 0.14,
        "usd_per_m_cached_in": null,
        "context_length": 16384,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 23.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.2786,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.8201,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 157.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.0624,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 54.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6853,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.3547,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 86.6642,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 50.3211,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 80.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.2836,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Mungert/Phi-4-mini-instruct.gguf:IQ2_XS",
      "family": "phi-4",
      "member": "mini",
      "quant": "IQ2_XS",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "Mungert/Phi-4-mini-instruct.gguf",
      "publisher": "mungert",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W37"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 4.41,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-11T04:13:11.031Z",
          "last_at": "2026-09-11T04:13:11.031Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:BF16",
      "family": "functiongemma",
      "member": "270m~mungert",
      "quant": "BF16",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "Mungert/functiongemma-270m-it-GGUF",
      "publisher": "Mungert",
      "derivative": true,
      "cohort": 3,
      "rung": 4,
      "disk_mib": 517,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Mungert/functiongemma-270m-it-GGUF",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 1.5357
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 397.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1797,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 20.08,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.619,
          "first_at": "2026-10-25T05:12:21.592Z",
          "last_at": "2026-10-25T05:12:21.592Z",
          "caps": [
            "lang.bn",
            "lang.en",
            "lang.hi",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.5357,
          "usd_per_m_energy": 0.0727,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2636.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1809,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 912.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0338,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 369.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3028,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.7374,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 25.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7554,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1338.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1917,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:IQ4_NL",
      "family": "functiongemma",
      "member": "270m~mungert",
      "quant": "IQ4_NL",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "Mungert/functiongemma-270m-it-GGUF",
      "publisher": "Mungert",
      "derivative": true,
      "cohort": 3,
      "rung": 9,
      "disk_mib": 150,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Mungert/functiongemma-270m-it-GGUF",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.7589
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1369.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0519,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 40.24,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.268,
          "first_at": "2026-10-25T02:00:06.792Z",
          "last_at": "2026-10-25T02:00:06.792Z",
          "caps": [
            "lang.bn",
            "lang.hi",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.7589,
          "usd_per_m_energy": 0.063,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9087.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0525,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3144.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0097,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1274.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0876,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 51.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.3688,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 89.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7962,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4614.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0556,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q3_K_S",
      "family": "functiongemma",
      "member": "270m~mungert",
      "quant": "Q3_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "Mungert/functiongemma-270m-it-GGUF",
      "publisher": "Mungert",
      "derivative": true,
      "cohort": 3,
      "rung": 2,
      "disk_mib": 179,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Mungert/functiongemma-270m-it-GGUF",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.8877
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1147.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.062,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 34.43,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.022,
          "first_at": "2026-10-25T02:11:01.586Z",
          "last_at": "2026-10-25T02:11:01.586Z",
          "caps": [
            "lang.hi",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.8877,
          "usd_per_m_energy": 0.0562,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7615.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0626,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2635.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0116,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1067.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1046,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6341,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 74.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9504,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3867.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0663,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q4_1",
      "family": "functiongemma",
      "member": "270m~mungert",
      "quant": "Q4_1",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "Mungert/functiongemma-270m-it-GGUF",
      "publisher": "Mungert",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 176,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Mungert/functiongemma-270m-it-GGUF",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.912
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1166.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0609,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 33.51,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.682,
          "first_at": "2026-10-25T02:06:29.750Z",
          "last_at": "2026-10-25T02:06:29.750Z",
          "caps": [
            "lang.bn",
            "lang.en",
            "lang.hi",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.912,
          "usd_per_m_energy": 0.0467,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7745.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0615,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2680.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0114,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1086.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1028,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 44.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6068,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 76.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9345,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3932.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0652,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q4_K_M",
      "family": "functiongemma",
      "member": "270m~mungert",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "Mungert/functiongemma-270m-it-GGUF",
      "publisher": "Mungert",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 241,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Mungert/functiongemma-270m-it-GGUF",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9347
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 852.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0835,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 32.75,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.059,
          "first_at": "2026-10-25T03:08:10.670Z",
          "last_at": "2026-10-25T03:08:10.670Z",
          "caps": [
            "lang.bn",
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.9347,
          "usd_per_m_energy": 0.0572,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5656.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0843,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1957.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0156,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 793.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1409,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 32.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.2013,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 55.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2805,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2872.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0893,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Mungert/functiongemma-270m-it-GGUF:Q8_0",
      "family": "functiongemma",
      "member": "270m~mungert",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "Mungert/functiongemma-270m-it-GGUF",
      "publisher": "Mungert",
      "derivative": true,
      "cohort": 3,
      "rung": 1,
      "disk_mib": 462,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Mungert/functiongemma-270m-it-GGUF",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 1.3784
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 444.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1605,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 22.34,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.446,
          "first_at": "2026-10-25T04:03:11.909Z",
          "last_at": "2026-10-25T04:03:11.909Z",
          "caps": [
            "lang.en",
            "lang.hi",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.3784,
          "usd_per_m_energy": 0.0679,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2950.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1616,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1020.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0302,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 413.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2705,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.2318,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 28.99,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.4611,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1498.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1713,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q2_K",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "North-ML1/Forge-1-Mini-GGUF",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/North-ML1/Forge-1-Mini-GGUF",
        "fetched_at": "2026-11-05T08:38:52.300Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 1074.92,
          "effective_ctx_best": null,
          "joules_per_token_best": 0.028,
          "first_at": "2026-10-26T07:47:15.317Z",
          "last_at": "2026-10-26T07:47:15.317Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/North-ML1/Forge-1-Mini-GGUF:Q3_K_M",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "North-ML1/Forge-1-Mini-GGUF",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/North-ML1/Forge-1-Mini-GGUF",
        "fetched_at": "2026-11-05T08:38:52.300Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 628.82,
          "effective_ctx_best": null,
          "joules_per_token_best": 0.033,
          "first_at": "2026-10-25T19:51:34.864Z",
          "last_at": "2026-10-25T19:51:34.864Z",
          "caps": [
            "code.reason"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/QuantFactory/Llama-3.2-3B-Instruct-abliterated-GGUF:Q4_K_M",
      "family": "llama-3-2",
      "member": "3b~quantfactory-llama-3-2-abliterated",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-3B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-3B-Instruct",
      "publisher": "QuantFactory",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 1925,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 106.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6799,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 24.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2996,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 708.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6752,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 245.03,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1305,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 99.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1389,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.05,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 17.9108,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.4222,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 359.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.717,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/QuantFactory/OpenHathi-7B-Hi-v0.1-Base-GGUF:Q4_K_M",
      "family": "openhathi",
      "member": "7b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "sarvamai/OpenHathi-7B-Hi-v0.1-Base",
      "hub_repo": "sarvamai/OpenHathi-7B-Hi-v0.1-Base",
      "publisher": "QuantFactory",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 3979,
      "active_share": 1,
      "licence": {
        "id": "llama2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/sarvamai/OpenHathi-7B-Hi-v0.1-Base",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 1.1278
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 51.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4379,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.8258,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 342.59,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4004,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 29.84,
          "effective_ctx_best": null,
          "joules_per_token_best": 3.904,
          "first_at": "2026-10-20T08:42:13.165Z",
          "last_at": "2026-10-20T08:42:13.165Z",
          "caps": [],
          "usd_per_m_out": 1.1278,
          "usd_per_m_energy": 0.1084,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 50.85,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.273,
          "first_at": "2026-11-03T21:52:52.959Z",
          "last_at": "2026-11-03T21:52:52.959Z",
          "caps": [
            "code.reason"
          ],
          "usd_per_m_out": 2.2569,
          "usd_per_m_energy": 0.0631,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 37.8624,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.37,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 22.0209,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 173.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4917,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M",
      "family": "qwen-2-5",
      "member": "0.5b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen2.5-0.5B-Instruct",
      "hub_repo": "Qwen/Qwen2.5-0.5B-Instruct",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 468,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen2.5-0.5B-Instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W37",
        "2026-W38",
        "2026-W39",
        "2026-W40",
        "2026-W41",
        "2026-W42",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.1288
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 438.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1626,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 18.98,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 1.658,
          "first_at": "2026-10-26T05:13:45.047Z",
          "last_at": "2026-10-26T05:13:45.047Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 1.6226,
          "usd_per_m_energy": 0.0461,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2912.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1637,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 239.08,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 0.768,
          "first_at": "2026-10-19T22:01:17.067Z",
          "last_at": "2026-10-19T22:01:17.067Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 0.1288,
          "usd_per_m_energy": 0.0213,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 408.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.274,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.2855,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W37",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": 31.82,
          "effective_ctx_best": 16384,
          "joules_per_token_best": 2.277,
          "first_at": "2026-09-13T03:29:59.307Z",
          "last_at": "2026-11-14T10:24:33.050Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.ja",
            "lang.kn",
            "lang.nl",
            "lang.ru",
            "lang.zh",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 2.2424,
          "usd_per_m_energy": 0.0632,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W45"
          ],
          "tokens_per_sec_best": 247.91,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 7.78,
          "first_at": "2026-09-13T16:05:44.419Z",
          "last_at": "2026-11-08T13:24:21.415Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.nl",
            "lang.ru",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 1.0352,
          "usd_per_m_energy": 0.2161,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 5.75,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-16T09:37:16.913Z",
          "last_at": "2026-09-16T09:37:16.913Z",
          "caps": [
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "est_usd_per_m_out": 25.2353,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/Qwen/Qwen2.5-0.5B-Instruct-GGUF:q4_k_m",
      "family": "qwen-2-5",
      "member": "0.5b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen2.5-0.5B-Instruct",
      "hub_repo": "Qwen/Qwen2.5-0.5B-Instruct",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 468,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen2.5-0.5B-Instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W37",
        "2026-W38",
        "2026-W40",
        "2026-W41",
        "2026-W42",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.1288
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 438.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1626,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 18.98,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 1.658,
          "first_at": "2026-10-26T05:13:45.047Z",
          "last_at": "2026-10-26T05:13:45.047Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 1.6226,
          "usd_per_m_energy": 0.0461,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2912.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1637,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 239.08,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 0.768,
          "first_at": "2026-10-19T22:01:17.067Z",
          "last_at": "2026-10-19T22:01:17.067Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 0.1288,
          "usd_per_m_energy": 0.0213,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 319.04,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 0.485,
          "first_at": "2026-11-03T21:23:08.756Z",
          "last_at": "2026-11-06T08:15:29.210Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 0.3508,
          "usd_per_m_energy": 0.0135,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.2855,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W37",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": 39.65,
          "effective_ctx_best": 16384,
          "joules_per_token_best": 1.092,
          "first_at": "2026-09-13T15:40:19.376Z",
          "last_at": "2026-11-14T10:38:26.168Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.ja",
            "lang.nl",
            "lang.pt",
            "lang.ru",
            "lang.si",
            "lang.zh",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 1.7996,
          "usd_per_m_energy": 0.0303,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W45"
          ],
          "tokens_per_sec_best": 247.91,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 7.78,
          "first_at": "2026-09-13T16:18:47.703Z",
          "last_at": "2026-11-08T13:24:21.415Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.nl",
            "lang.ru",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 1.0352,
          "usd_per_m_energy": 0.2161,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 5.75,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-16T09:37:16.913Z",
          "last_at": "2026-09-16T09:37:16.913Z",
          "caps": [
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "est_usd_per_m_out": 25.2353,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/Qwen/Qwen3-0.6B-GGUF:Q8_0",
      "family": "qwen-3",
      "member": "0.6b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-0.6B",
      "hub_repo": "Qwen/Qwen3-0.6B",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 2,
      "rung": 1,
      "disk_mib": 609,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-0.6B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 2.4169
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 337.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2119,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 12.79,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 1.896,
          "first_at": "2026-10-25T20:27:10.905Z",
          "last_at": "2026-10-25T20:27:10.905Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call"
          ],
          "usd_per_m_out": 2.4169,
          "usd_per_m_energy": 0.0527,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2238.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2131,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 774.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0399,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 313.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3569,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.5878,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 21.99,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.25,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1136.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2259,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Qwen/Qwen3-14B-GGUF:Q4_K_M",
      "family": "qwen-3",
      "member": "14b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-14B",
      "hub_repo": "Qwen/Qwen3-14B",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 8584,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-14B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3-14b",
        "ids": [
          "qwen/qwen3-14b"
        ],
        "usd_per_m_in": 0.12,
        "usd_per_m_out": 0.24,
        "usd_per_m_cached_in": null,
        "context_length": 40960,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 23.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.2577,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.7758,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 158.8,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.0448,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 54.95,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6807,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.3217,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 85.668,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 49.973,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 80.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.2643,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Qwen/Qwen3-14B-GGUF:Q8_0",
      "family": "qwen-3",
      "member": "14b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-14B",
      "hub_repo": "Qwen/Qwen3-14B",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 2,
      "rung": 1,
      "disk_mib": 14971,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-14B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3-14b",
        "ids": [
          "qwen/qwen3-14b"
        ],
        "usd_per_m_in": 0.12,
        "usd_per_m_out": 0.24,
        "usd_per_m_cached_in": null,
        "context_length": 40960,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 2.7431,
        "vs_api": 11.43
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.0609,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 13.481,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 91.05,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.3675,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 15.53,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 13.517,
          "first_at": "2026-10-23T21:59:41.480Z",
          "last_at": "2026-10-23T21:59:41.480Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "usd_per_m_out": 2.7431,
          "usd_per_m_energy": 0.3755,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.6876,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 159.9147,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 93.4333,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 46.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.8052,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q4_K_M",
      "family": "qwen-3",
      "member": "4b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-4B",
      "hub_repo": "Qwen/Qwen3-4B",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 2381,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-4B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W37",
        "2026-W43",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.6414
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 86.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8452,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6258,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 572.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8357,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 50.44,
          "effective_ctx_best": 16384,
          "joules_per_token_best": 4.298,
          "first_at": "2026-10-20T09:07:06.225Z",
          "last_at": "2026-10-20T09:07:06.225Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.ta",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "usd_per_m_out": 0.6414,
          "usd_per_m_energy": 0.1194,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 49.84,
          "effective_ctx_best": 40960,
          "joules_per_token_best": 2.344,
          "first_at": "2026-11-06T23:36:56.826Z",
          "last_at": "2026-11-06T23:36:56.826Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "usd_per_m_out": 2.2766,
          "usd_per_m_energy": 0.0651,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 22.2966,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.63,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.9502,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 140.72,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-09-11T20:03:51.058Z",
          "last_at": "2026-09-11T20:03:51.058Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 1.8348,
          "usd_per_m_energy": 0.0444,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_0",
      "family": "qwen-3",
      "member": "4b",
      "quant": "Q5_0",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-4B",
      "hub_repo": "Qwen/Qwen3-4B",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 2692,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-4B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W38",
        "2026-W40",
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.6799
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W40"
          ],
          "tokens_per_sec_best": 22.66,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-10-01T06:52:22.001Z",
          "last_at": "2026-10-01T06:52:22.001Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.ml",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "usd_per_m_out": 3.2287,
          "usd_per_m_energy": 0.0429,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8527,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 506.37,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9454,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 47.96,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 3.857,
          "first_at": "2026-10-20T11:05:35.487Z",
          "last_at": "2026-10-20T11:05:35.487Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 0.6799,
          "usd_per_m_energy": 0.1071,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6017,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 25.3159,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.6914,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 11.16,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-16T18:23:12.120Z",
          "last_at": "2026-09-16T18:23:12.120Z",
          "caps": [
            "faithful_tool_result",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 23.1578,
          "usd_per_m_energy": 0.56,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q5_K_M",
      "family": "qwen-3",
      "member": "4b",
      "quant": "Q5_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-4B",
      "hub_repo": "Qwen/Qwen3-4B",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 1,
      "rung": 3,
      "disk_mib": 2755,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-4B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W38",
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.695
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 74.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9821,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8999,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 494.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9676,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 46.99,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 4.694,
          "first_at": "2026-10-20T11:13:56.510Z",
          "last_at": "2026-10-20T11:13:56.510Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 0.695,
          "usd_per_m_energy": 0.1304,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 69.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6398,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 25.8708,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 15.0647,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 35.37,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T16:20:53.068Z",
          "last_at": "2026-09-15T16:20:53.068Z",
          "caps": [
            "faithful_tool_result",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 7.3082,
          "usd_per_m_energy": 0.1767,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/Qwen/Qwen3-4B-GGUF:Q8_0",
      "family": "qwen-3",
      "member": "4b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-4B",
      "hub_repo": "Qwen/Qwen3-4B",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 1,
      "rung": 1,
      "disk_mib": 4082,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-4B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W38",
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.8682
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 50.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4767,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.9085,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 333.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4369,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 38.86,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 5.213,
          "first_at": "2026-10-20T11:46:02.910Z",
          "last_at": "2026-10-20T11:46:02.910Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "usd_per_m_out": 0.8682,
          "usd_per_m_energy": 0.1448,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 46.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.4525,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 38.8975,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 22.6507,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 17.06,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T17:38:02.239Z",
          "last_at": "2026-09-15T17:38:02.239Z",
          "caps": [
            "faithful_tool_result",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "usd_per_m_out": 15.2152,
          "usd_per_m_energy": 0.3664,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/Qwen/Qwen3-VL-2B-Instruct-GGUF:Q4_K_M",
      "family": "qwen-3",
      "member": "vl-2b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-VL-2B-Instruct",
      "hub_repo": "Qwen/Qwen3-VL-2B-Instruct",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 1056,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-VL-2B-Instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 3.3849
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 194.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3694,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 9.24,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 5.193,
          "first_at": "2026-11-04T04:43:49.469Z",
          "last_at": "2026-11-04T04:43:49.469Z",
          "caps": [
            "faithful_tool_result",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 3.3849,
          "usd_per_m_energy": 0.1442,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1290.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3698,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 446.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.07,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 181.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6209,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.7333,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.665,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 655.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3922,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Qwen/Qwen3-VL-4B-Instruct-GGUF:F16",
      "family": "qwen-3",
      "member": "vl-4b",
      "quant": "F16",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-VL-4B-Instruct",
      "hub_repo": "Qwen/Qwen3-VL-4B-Instruct",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 2,
      "rung": 4,
      "disk_mib": 7678,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-VL-4B-Instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 3.0785
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 26.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.8868,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.9425,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 177.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7193,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 11.91,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 17.505,
          "first_at": "2026-10-21T21:52:57.064Z",
          "last_at": "2026-10-21T21:52:57.064Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 3.0785,
          "usd_per_m_energy": 0.4863,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 24.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.7319,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 76.456,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 44.3797,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 90.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.9118,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Qwen/Qwen3-VL-8B-Instruct-GGUF:F16",
      "family": "qwen-3",
      "member": "vl-8b",
      "quant": "F16",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-VL-8B-Instruct",
      "hub_repo": "Qwen/Qwen3-VL-8B-Instruct",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 2,
      "rung": 4,
      "disk_mib": 15628,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-VL-8B-Instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3-vl-8b-instruct",
        "ids": [
          "qwen/qwen3-vl-8b-instruct"
        ],
        "usd_per_m_in": 0.25,
        "usd_per_m_out": 0.75,
        "usd_per_m_cached_in": 0.12,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 12.0826,
        "vs_api": 16.11
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.3691,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.03,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.2359,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 87.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.6087,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 3.57,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 34.229,
          "first_at": "2026-10-23T22:32:04.290Z",
          "last_at": "2026-10-23T22:32:04.290Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.pa",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 12.0826,
          "usd_per_m_energy": 0.9508,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.1591,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 167.3806,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 97.3143,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 44.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.0729,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Qwen/Qwen3-VL-8B-Instruct-GGUF:Q4_K_M",
      "family": "qwen-3",
      "member": "vl-8b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-VL-8B-Instruct",
      "hub_repo": "Qwen/Qwen3-VL-8B-Instruct",
      "publisher": "Qwen",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 4794,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-VL-8B-Instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3-vl-8b-instruct",
        "ids": [
          "qwen/qwen3-vl-8b-instruct"
        ],
        "usd_per_m_in": 0.25,
        "usd_per_m_out": 0.75,
        "usd_per_m_cached_in": 0.12,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 42.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7477,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.4735,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 284.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6896,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 98.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3488,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 39.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.8951,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.63,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 45.9347,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.8364,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 144.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8017,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/Reza2kn/Qianfan-OCR-GGUF:Q4_K_M",
      "family": "qianfan-ocr",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "baidu/Qianfan-OCR",
      "hub_repo": "baidu/Qianfan-OCR",
      "publisher": "Reza2kn",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 2596,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/baidu/Qianfan-OCR",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 6.4715
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 79.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9238,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 18.25,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7824,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 525.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9115,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 181.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.179,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 17.56,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 5.108,
          "first_at": "2026-11-06T12:38:36.372Z",
          "last_at": "2026-11-06T12:38:36.372Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 6.4715,
          "usd_per_m_energy": 0.1419,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 24.3616,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.1637,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 266.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.969,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/TeichAI/Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill-GGUF:IQ3_XS",
      "family": "deepseek-v3-2",
      "member": "devstral-small-2505-speciale-distill~teichai",
      "quant": "IQ3_XS",
      "engine": "ollama",
      "modality": "text",
      "base": "TeichAI/Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill",
      "hub_repo": "TeichAI/Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill",
      "publisher": "TeichAI",
      "derivative": true,
      "cohort": 3,
      "rung": 2,
      "disk_mib": 9448,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/TeichAI/Devstral-Small-2505-Deepseek-V3.2-Speciale-Distill",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 105.8485
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 21.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.6183,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 0.36,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 98.686,
          "first_at": "2026-11-08T00:47:15.737Z",
          "last_at": "2026-11-08T00:47:15.737Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 105.8485,
          "usd_per_m_energy": 2.7413,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 144.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.3561,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 49.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7633,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 20.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.8931,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 95.9281,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 55.3951,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 73.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.6028,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/HuggingFaceTB_SmolLM3-3B-GGUF:Q4_K_M",
      "family": "smollm-3",
      "member": "3b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "HuggingFaceTB/SmolLM3-3B",
      "hub_repo": "HuggingFaceTB/SmolLM3-3B",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 1826,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/HuggingFaceTB/SmolLM3-3B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 1.4566
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 112.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6442,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 25.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2299,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 746.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6403,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 258.32,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1235,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 77.59,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 1.432,
          "first_at": "2026-11-06T22:34:56.010Z",
          "last_at": "2026-11-06T22:34:56.010Z",
          "caps": [
            "code.reason",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 1.4566,
          "usd_per_m_energy": 0.0398,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 16.9691,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.8717,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 379.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6799,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:F16",
      "family": "llama-3-2",
      "member": "1b",
      "quant": "F16",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-1B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-1B-Instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 2,
      "rung": 4,
      "disk_mib": 2364,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-1B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W35",
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-1b-instruct",
        "ids": [
          "meta-llama/llama-3.2-1b-instruct"
        ],
        "usd_per_m_in": 0.027,
        "usd_per_m_out": 0.201,
        "usd_per_m_cached_in": null,
        "context_length": 60000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 7.521,
        "vs_api": 37.42
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 86.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.839,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 4.3,
          "effective_ctx_best": null,
          "joules_per_token_best": 5.507,
          "first_at": "2026-10-26T06:16:09.440Z",
          "last_at": "2026-10-26T06:16:09.440Z",
          "caps": [
            "faithful_tool_result",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 7.521,
          "usd_per_m_energy": 0.153,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 576.63,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8297,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 199.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1621,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 80.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4031,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 6.72,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-27T05:29:13.386Z",
          "last_at": "2026-08-27T05:29:13.386Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.nl",
            "lang.te",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "verdict": "proved"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 22.0897,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.8564,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 292.81,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8817,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:IQ3_M",
      "family": "llama-3-2",
      "member": "1b",
      "quant": "IQ3_M",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-1B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-1B-Instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 2,
      "rung": 2,
      "disk_mib": 626,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-1B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-1b-instruct",
        "ids": [
          "meta-llama/llama-3.2-1b-instruct"
        ],
        "usd_per_m_in": 0.027,
        "usd_per_m_out": 0.201,
        "usd_per_m_cached_in": null,
        "context_length": 60000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 5.8572,
        "vs_api": 29.14
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 328.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2179,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 5.28,
          "effective_ctx_best": null,
          "joules_per_token_best": 6.499,
          "first_at": "2026-10-25T21:51:30.809Z",
          "last_at": "2026-10-25T21:51:30.809Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 5.8572,
          "usd_per_m_energy": 0.1805,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2177.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2191,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 753.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.041,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 305.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3669,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.7461,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 21.4,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.3403,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1105.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2322,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q3_K_L",
      "family": "llama-3-2",
      "member": "1b",
      "quant": "Q3_K_L",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-1B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-1B-Instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 2,
      "rung": 9,
      "disk_mib": 698,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-1B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-1b-instruct",
        "ids": [
          "meta-llama/llama-3.2-1b-instruct"
        ],
        "usd_per_m_in": 0.027,
        "usd_per_m_out": 0.201,
        "usd_per_m_cached_in": null,
        "context_length": 60000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 3.0082,
        "vs_api": 14.97
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 294.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2431,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 10.3,
          "effective_ctx_best": null,
          "joules_per_token_best": 4.301,
          "first_at": "2026-10-25T22:08:52.934Z",
          "last_at": "2026-10-25T22:08:52.934Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.kn",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 3.0082,
          "usd_per_m_energy": 0.1195,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1952.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2443,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 675.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0459,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 273.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4093,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.4104,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.728,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 991.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.259,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q4_K_M",
      "family": "llama-3-2",
      "member": "1b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-1B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-1B-Instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 770,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-1B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W35",
        "2026-W37",
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-1b-instruct",
        "ids": [
          "meta-llama/llama-3.2-1b-instruct"
        ],
        "usd_per_m_in": 0.027,
        "usd_per_m_out": 0.201,
        "usd_per_m_cached_in": null,
        "context_length": 60000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 2.5362,
        "vs_api": 12.62
      },
      "classes": {
        "darwin/amd64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 4.73,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-08T15:08:36.906Z",
          "last_at": "2026-09-08T15:08:36.906Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "est_usd_per_m_out": 8.7066,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 266.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2684,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 12.24,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 3.229,
          "first_at": "2026-10-26T05:49:49.813Z",
          "last_at": "2026-10-26T05:49:49.813Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 2.5362,
          "usd_per_m_energy": 0.0897,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1770.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2695,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 612.59,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0507,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 248.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4518,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 103.86,
          "effective_ctx_best": 8192,
          "joules_per_token_best": null,
          "first_at": "2026-08-26T09:19:24.675Z",
          "last_at": "2026-08-26T09:19:24.675Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.ml",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "verdict": "proved"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.075,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.1172,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 11.62,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-29T23:25:45.622Z",
          "last_at": "2026-08-29T23:25:45.622Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.nl",
            "lang.te",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "est_usd_per_m_out": 3.5441,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 898.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2857,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/Llama-3.2-1B-Instruct-GGUF:Q8_0",
      "family": "llama-3-2",
      "member": "1b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-1B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-1B-Instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 2,
      "rung": 1,
      "disk_mib": 1259,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-1B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-1b-instruct",
        "ids": [
          "meta-llama/llama-3.2-1b-instruct"
        ],
        "usd_per_m_in": 0.027,
        "usd_per_m_out": 0.201,
        "usd_per_m_cached_in": null,
        "context_length": 60000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 3.7165,
        "vs_api": 18.49
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 163.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4414,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 8.46,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 3.263,
          "first_at": "2026-10-26T02:33:45.356Z",
          "last_at": "2026-10-26T02:33:45.356Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 3.7165,
          "usd_per_m_energy": 0.0906,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1082.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4411,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 374.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0839,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 151.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7414,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.6311,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.7666,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 549.8,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4679,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:F16",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "F16",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "bartowski/Llama-3.2-3B-Instruct-GGUF",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 38.31,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 11.625,
          "first_at": "2026-10-21T22:57:27.099Z",
          "last_at": "2026-10-21T22:57:27.099Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.ml",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:IQ3_M",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "IQ3_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "bartowski/Llama-3.2-3B-Instruct-GGUF",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 71.97,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 8.663,
          "first_at": "2026-10-20T12:02:23.958Z",
          "last_at": "2026-10-20T12:02:23.958Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q3_K_L",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "Q3_K_L",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "bartowski/Llama-3.2-3B-Instruct-GGUF",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 78.82,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 8.999,
          "first_at": "2026-10-20T12:17:42.989Z",
          "last_at": "2026-10-20T12:17:42.989Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "lang.en",
            "lang.kn",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "Q4_K_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "bartowski/Llama-3.2-3B-Instruct-GGUF",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 86.87,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 7.485,
          "first_at": "2026-10-20T09:16:07.616Z",
          "last_at": "2026-10-20T09:16:07.616Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 99.3,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 6.261,
          "first_at": "2026-11-08T11:39:02.516Z",
          "last_at": "2026-11-08T11:39:02.516Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q8_0",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "Q8_0",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "bartowski/Llama-3.2-3B-Instruct-GGUF",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 62.75,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 8.087,
          "first_at": "2026-10-21T22:29:12.561Z",
          "last_at": "2026-10-21T22:29:12.561Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "lang.en",
            "lang.kn",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/Llama-3.2-3B-Instruct-uncensored-GGUF:IQ3_XS",
      "family": "llama-3-2",
      "member": "3b~chuanli11-llama-3-2-uncensored",
      "quant": "IQ3_XS",
      "engine": "ollama",
      "modality": "text",
      "base": "chuanli11/Llama-3.2-3B-Instruct-uncensored",
      "hub_repo": "chuanli11/Llama-3.2-3B-Instruct-uncensored",
      "publisher": "bartowski",
      "derivative": true,
      "cohort": 3,
      "rung": 2,
      "disk_mib": 1569,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/chuanli11/Llama-3.2-3B-Instruct-uncensored",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 130.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.552,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 30.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0498,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 868.8,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.55,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 300.63,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1054,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 121.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.926,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 46.8,
          "effective_ctx_best": 8192,
          "joules_per_token_best": null,
          "first_at": "2026-08-26T10:20:01.128Z",
          "last_at": "2026-08-26T10:20:01.128Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.97,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.537,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.4601,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 441.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5837,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:BF16",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "BF16",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "bartowski/google_gemma-4-E2B-it-GGUF",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 28.24,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 3.553,
          "first_at": "2026-10-23T14:27:05.533Z",
          "last_at": "2026-10-23T14:27:05.533Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:IQ2_M",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "IQ2_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "bartowski/google_gemma-4-E2B-it-GGUF",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 71.23,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.515,
          "first_at": "2026-10-23T16:08:01.969Z",
          "last_at": "2026-10-23T16:08:01.969Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:IQ3_XS",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "IQ3_XS",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "bartowski/google_gemma-4-E2B-it-GGUF",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 59.02,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.785,
          "first_at": "2026-10-23T16:16:48.547Z",
          "last_at": "2026-10-23T16:16:48.547Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:Q3_K_L",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "Q3_K_L",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "bartowski/google_gemma-4-E2B-it-GGUF",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 73,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.561,
          "first_at": "2026-10-23T16:24:05.055Z",
          "last_at": "2026-10-23T16:24:05.055Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:Q4_0",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "Q4_0",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "bartowski/google_gemma-4-E2B-it-GGUF",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 82.25,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.362,
          "first_at": "2026-10-23T16:31:33.170Z",
          "last_at": "2026-10-23T16:31:33.170Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:Q4_K_M",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "Q4_K_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "bartowski/google_gemma-4-E2B-it-GGUF",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 73.42,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.438,
          "first_at": "2026-10-23T16:38:22.890Z",
          "last_at": "2026-10-23T16:38:22.890Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/google_gemma-4-E2B-it-GGUF:Q8_0",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "Q8_0",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "bartowski/google_gemma-4-E2B-it-GGUF",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 58.97,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.696,
          "first_at": "2026-10-23T16:46:09.915Z",
          "last_at": "2026-10-23T16:46:09.915Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/krutrim-ai-labs_Krutrim-2-instruct-GGUF:Q4_K_M",
      "family": "krutrim",
      "member": "2",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "krutrim-ai-labs/Krutrim-2-instruct",
      "hub_repo": "krutrim-ai-labs/Krutrim-2-instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 7130,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "krutrim-community-license-agreement-version-1.0",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/krutrim-ai-labs/Krutrim-2-instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 2.671
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 28.8,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.6658,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.4547,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 191.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.5228,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 13.56,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 30.182,
          "first_at": "2026-10-20T08:52:49.634Z",
          "last_at": "2026-10-20T08:52:49.634Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 2.671,
          "usd_per_m_energy": 0.8384,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 14.49,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 22.429,
          "first_at": "2026-11-06T17:55:41.851Z",
          "last_at": "2026-11-06T17:55:41.851Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 8.0973,
          "usd_per_m_energy": 0.623,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.09,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 70.4354,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 40.8376,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 97.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.6993,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ2_M",
      "family": "phi-4",
      "member": "mini",
      "quant": "IQ2_M",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 2121,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W35",
        "2026-W42"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "darwin/arm64/apple/16g",
        "usd_per_m_out": 2.6836
      },
      "classes": {
        "darwin/amd64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 1.06,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-28T06:09:23.417Z",
          "last_at": "2026-08-28T06:09:23.417Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "est_usd_per_m_out": 39.8883,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W42"
          ],
          "tokens_per_sec_best": 27.09,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-10-16T05:20:46.408Z",
          "last_at": "2026-10-16T05:20:46.408Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 2.6836,
          "usd_per_m_energy": 0.0359,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4394,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 642.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7441,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 222.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1445,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 90.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2567,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 37.31,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-26T05:02:24.459Z",
          "last_at": "2026-08-26T05:02:24.459Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19.8088,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.5211,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 326.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7905,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ3_XXS",
      "family": "phi-4",
      "member": "mini",
      "quant": "IQ3_XXS",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 2121,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W42"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "darwin/arm64/apple/16g",
        "usd_per_m_out": 105.3598
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W42"
          ],
          "tokens_per_sec_best": 0.69,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-10-16T04:10:16.563Z",
          "last_at": "2026-10-16T04:10:16.563Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 105.3598,
          "usd_per_m_energy": 1.409,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4394,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 642.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7441,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 222.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1445,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 90.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2567,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19.8088,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.5211,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 326.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7905,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:IQ4_XS",
      "family": "phi-4",
      "member": "mini",
      "quant": "IQ4_XS",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 2121,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W42"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "darwin/arm64/apple/16g",
        "usd_per_m_out": 2.0082
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W42"
          ],
          "tokens_per_sec_best": 36.2,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-10-16T05:29:02.559Z",
          "last_at": "2026-10-16T05:29:02.559Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 2.0082,
          "usd_per_m_energy": 0.0269,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4394,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 642.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7441,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 222.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1445,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 90.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2567,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19.8088,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.5211,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 326.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7905,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q3_K_L",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q3_K_L",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 2121,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 96.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7508,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4394,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 642.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7441,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 222.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1445,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 90.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2567,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 32.06,
          "effective_ctx_best": 8192,
          "joules_per_token_best": null,
          "first_at": "2026-08-26T09:30:19.133Z",
          "last_at": "2026-08-26T09:30:19.133Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19.8088,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.5211,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 326.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7905,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q3_K_S",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q3_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 2121,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 96.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7508,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "darwin/arm64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 28.18,
          "effective_ctx_best": 8192,
          "joules_per_token_best": null,
          "first_at": "2026-08-28T09:05:05.960Z",
          "last_at": "2026-08-28T09:05:05.960Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "est_usd_per_m_out": 1.5004,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4394,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 642.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7441,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 222.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1445,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 90.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2567,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19.8088,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.5211,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 326.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7905,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q4_0",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q4_0",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 2121,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W42"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "darwin/arm64/apple/16g",
        "usd_per_m_out": 2.1547
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W42"
          ],
          "tokens_per_sec_best": 33.74,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-10-16T05:48:46.611Z",
          "last_at": "2026-10-16T05:48:46.611Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 2.1547,
          "usd_per_m_energy": 0.0288,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4394,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 642.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7441,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 222.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1445,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 90.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2567,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19.8088,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.5211,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 326.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7905,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/microsoft_Phi-4-mini-instruct-GGUF:Q4_K_M",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 2121,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W42"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "darwin/arm64/apple/16g",
        "usd_per_m_out": 2.1572
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W42"
          ],
          "tokens_per_sec_best": 33.7,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-10-16T05:58:24.228Z",
          "last_at": "2026-10-16T05:58:24.228Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 2.1572,
          "usd_per_m_energy": 0.0288,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4394,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 642.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7441,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 222.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1445,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 90.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2567,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19.8088,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.5211,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 326.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7905,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/moonshotai_Kimi-Linear-48B-A3B-Instruct-GGUF:Q4_K_M",
      "family": "kimi-linear",
      "member": "48b-a3b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "moonshotai/Kimi-Linear-48B-A3B-Instruct",
      "hub_repo": "moonshotai/Kimi-Linear-48B-A3B-Instruct",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 28668,
      "active_share": 0.0625,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-Linear-48B-A3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 11.174
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 114.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8227,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 4.81,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 89.44,
          "first_at": "2026-10-27T17:06:10.540Z",
          "last_at": "2026-10-27T18:18:44.290Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 11.174,
          "usd_per_m_energy": 2.4844,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 760.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.657,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 263.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2042,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 106.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2641,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 21.6787,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.6073,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 386.32,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7237,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/openai_gpt-oss-20b-GGUF-MXFP4-Experimental:MXFP4",
      "family": "gpt-oss",
      "member": "20b~bartowski-openai-experimental",
      "quant": "MXFP4",
      "engine": "ollama",
      "modality": "text",
      "base": "openai/gpt-oss-20b",
      "hub_repo": "openai/gpt-oss-20b",
      "publisher": "bartowski",
      "derivative": true,
      "cohort": 3,
      "rung": 9,
      "disk_mib": 11548,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/gpt-oss-20b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "openai/gpt-oss-20b",
        "ids": [
          "openai/gpt-oss-20b",
          "openai/gpt-oss-20b:batch"
        ],
        "usd_per_m_in": 0.03,
        "usd_per_m_out": 0.13,
        "usd_per_m_cached_in": 0.03,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.112,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 9.2163,
        "vs_api": 70.89
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.5202,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 4.32,
          "effective_ctx_best": null,
          "joules_per_token_best": 23.051,
          "first_at": "2026-10-28T13:02:50.935Z",
          "last_at": "2026-10-28T13:02:50.935Z",
          "caps": [],
          "usd_per_m_out": 9.2163,
          "usd_per_m_energy": 0.6403,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 118.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.1166,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 40.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9747,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.3067,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 119.9552,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 69.2845,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 59.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.4319,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/sarvam-1-GGUF:Q4_K_M",
      "family": "sarvam",
      "member": "1",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "sarvamai/sarvam-1",
      "hub_repo": "sarvamai/sarvam-1",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 1476,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/sarvamai/sarvam-1",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W37",
        "2026-W38",
        "2026-W42",
        "2026-W43",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.2843
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W42"
          ],
          "tokens_per_sec_best": 62.17,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-16T07:42:56.645Z",
          "last_at": "2026-10-16T07:42:56.645Z",
          "caps": [
            "lang.bn",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.pa",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.1609,
          "usd_per_m_energy": 0.0156,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 32.09,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9853,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 923.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5173,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 111.22,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 4.472,
          "first_at": "2026-10-19T21:26:34.573Z",
          "last_at": "2026-10-19T21:26:34.573Z",
          "caps": [
            "lang.bn",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.2843,
          "usd_per_m_energy": 0.1242,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 121.44,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 1.494,
          "first_at": "2026-11-06T09:32:43.016Z",
          "last_at": "2026-11-06T09:32:43.016Z",
          "caps": [
            "lang.bn",
            "lang.gu",
            "lang.hi",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.9283,
          "usd_per_m_energy": 0.0415,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 13.6692,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W37",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": 12.97,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 7.576,
          "first_at": "2026-09-13T04:59:54.003Z",
          "last_at": "2026-11-14T05:22:23.150Z",
          "caps": [
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.pa",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 5.5646,
          "usd_per_m_energy": 0.2104,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38",
            "2026-W45"
          ],
          "tokens_per_sec_best": 178.24,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 4.453,
          "first_at": "2026-09-14T04:43:21.962Z",
          "last_at": "2026-11-08T13:24:58.945Z",
          "caps": [
            "instruct.follow",
            "lang.bn",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.4444,
          "usd_per_m_energy": 0.1237,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 23.48,
          "effective_ctx_best": 8192,
          "joules_per_token_best": null,
          "first_at": "2026-09-16T11:05:51.485Z",
          "last_at": "2026-09-16T11:05:51.485Z",
          "caps": [
            "lang.gu",
            "lang.pa",
            "output.schema",
            "structured_output"
          ],
          "est_usd_per_m_out": 6.2148,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/bartowski/sarvamai_sarvam-m-GGUF:Q4_K_M",
      "family": "sarvam",
      "member": "m",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "sarvamai/sarvam-m",
      "hub_repo": "sarvamai/sarvam-m",
      "publisher": "bartowski",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 13669,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/sarvamai/sarvam-m",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 2.6611
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.4658,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.9713,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 99.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.8897,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 15.61,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 12.472,
          "first_at": "2026-10-23T17:57:25.304Z",
          "last_at": "2026-10-23T17:57:25.304Z",
          "caps": [
            "code.reason",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "reason.channel",
            "research.synthesize",
            "thinking"
          ],
          "usd_per_m_out": 2.6611,
          "usd_per_m_energy": 0.3464,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.7734,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 144.0282,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 83.7715,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 50.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.2799,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bluevoid-pl/Muse-Glimmer-30B-pruned-V2.5-en-pl-GGUF:Q4_K_M",
      "family": "muse-glimmer",
      "member": "30b~bluevoid-pl-pruned-v2-5-en",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-models/Muse-Glimmer-30B",
      "hub_repo": "meta-models/Muse-Glimmer-30B",
      "publisher": "bluevoid-pl",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 15585,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/meta-models/Muse-Glimmer-30B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta/muse-glimmer-30b",
        "ids": [
          "meta/muse-glimmer-30b",
          "meta/muse-glimmer-30b:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.04,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.96,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.3471,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.1776,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 87.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.5929,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 30.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4238,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.1314,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 167.3107,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 97.2736,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 44.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.0557,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/bullerwins/translategemma-4b-it-GGUF:Q4_K_M",
      "family": "translategemma",
      "member": "4b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/translategemma-4b-it",
      "hub_repo": "google/translategemma-4b-it",
      "publisher": "bullerwins",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 2374,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/translategemma-4b-it",
        "fetched_at": "2026-11-10T15:57:15.731Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 1.4632
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 86.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8427,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.95,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6215,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 574.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8333,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 198.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1628,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 77.54,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 4.092,
          "first_at": "2026-11-06T22:01:09.469Z",
          "last_at": "2026-11-06T22:01:09.469Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.pa",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.4632,
          "usd_per_m_energy": 0.1137,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 22.2269,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.9263,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 291.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8855,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/cduk/embeddinggemma-300m-GGUF-with-dense-modules:Q6_K",
      "family": "embeddinggemma",
      "member": "300m~cduk-embeddinggemma-with-dense-modules",
      "quant": "Q6_K",
      "engine": "ollama",
      "modality": "text",
      "base": "google/embeddinggemma-300m",
      "hub_repo": "google/embeddinggemma-300m",
      "publisher": "cduk",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 252,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/embeddinggemma-300m",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 814.97,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0873,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:35:41.959Z",
          "last_at": "2026-10-25T20:35:41.959Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 187.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1629,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5409.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0881,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1871.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0164,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 758.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1473,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 30.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.3027,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 53.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.3392,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2746.81,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0934,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/cduk/embeddinggemma-300m-GGUF-with-dense-modules:Q8_0",
      "family": "embeddinggemma",
      "member": "300m~cduk-embeddinggemma-with-dense-modules",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/embeddinggemma-300m",
      "hub_repo": "google/embeddinggemma-300m",
      "publisher": "cduk",
      "derivative": true,
      "cohort": 3,
      "rung": 1,
      "disk_mib": 318,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/embeddinggemma-300m",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 645.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1103,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:36:38.141Z",
          "last_at": "2026-10-25T20:36:38.141Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 148.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2059,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4286.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1112,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1483.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0207,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 601.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.186,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 24.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.9074,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 42.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6911,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2176.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1178,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/cloudbjorn/Mistral-Medium-3.5-128B-Eschaton-Uncensored-GGUF:Q4_K_M",
      "family": "mistral-medium-3-5",
      "member": "128b~cloudbjorn-eschaton-uncensored",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "mistralai/Mistral-Medium-3.5-128B",
      "hub_repo": "mistralai/Mistral-Medium-3.5-128B",
      "publisher": "cloudbjorn",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 71427,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Medium-3.5-128B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-medium-3-5",
        "ids": [
          "mistralai/mistral-medium-3-5",
          "mistralai/mistral-medium-3-5:batch"
        ],
        "usd_per_m_in": 1.5,
        "usd_per_m_out": 7.5,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": 3.75,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 44.8263,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 134.1577,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 28.0218,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 13.4158,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 63.3042,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1173.6333,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 679.4719,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 32.4435,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/gaianet/embeddinggemma-300m-GGUF:Q2_K",
      "family": "embeddinggemma",
      "member": "300m",
      "quant": "Q2_K",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "gaianet/embeddinggemma-300m-GGUF",
      "publisher": "gaianet",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:31:59.132Z",
          "last_at": "2026-10-25T20:31:59.132Z",
          "caps": [
            "embed.vector"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/gaianet/embeddinggemma-300m-GGUF:Q3_K_S",
      "family": "embeddinggemma",
      "member": "300m",
      "quant": "Q3_K_S",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "gaianet/embeddinggemma-300m-GGUF",
      "publisher": "gaianet",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:32:24.192Z",
          "last_at": "2026-10-25T20:32:24.192Z",
          "caps": [
            "embed.vector"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/gaianet/embeddinggemma-300m-GGUF:Q4_K_M",
      "family": "embeddinggemma",
      "member": "300m",
      "quant": "Q4_K_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "gaianet/embeddinggemma-300m-GGUF",
      "publisher": "gaianet",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43",
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T04:18:19.570Z",
          "last_at": "2026-10-26T04:18:19.570Z",
          "caps": [
            "embed.vector"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/amd/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-28T13:19:30.256Z",
          "last_at": "2026-10-28T13:19:30.256Z",
          "caps": [
            "embed.vector"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43",
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T13:15:28.478Z",
          "last_at": "2026-10-27T19:17:13.006Z",
          "caps": [
            "embed.vector"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/ggml-org/GLM-4.6V-GGUF:Q4_K_M",
      "family": "glm-4-6",
      "member": "v",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "zai-org/GLM-4.6V",
      "hub_repo": "zai-org/GLM-4.6V",
      "publisher": "ggml-org",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 67184,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.6V",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.6v",
        "ids": [
          "z-ai/glm-4.6v"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.055,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 41.061,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 119.8467,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 20.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.1805,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.1212,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 58.3166,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1047.0558,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 628.2335,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 30.1868,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/ggml-org/embeddinggemma-300M-GGUF:Q8_0",
      "family": "embeddinggemma",
      "member": "300m",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/embeddinggemma-300m",
      "hub_repo": "google/embeddinggemma-300m",
      "publisher": "ggml-org",
      "derivative": false,
      "cohort": 1,
      "rung": 1,
      "disk_mib": 318,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/embeddinggemma-300m",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 645.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1103,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:33:16.812Z",
          "last_at": "2026-10-25T20:33:16.812Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 148.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2059,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4286.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1112,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1483.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0207,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 601.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.186,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 24.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.9074,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 42.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6911,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2176.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1178,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/ggml-org/gemma-3-1b-it-GGUF:Q4_K_M",
      "family": "gemma-3",
      "member": "1b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-3-1b-it",
      "hub_repo": "google/gemma-3-1b-it",
      "publisher": "ggml-org",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 768,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-1b-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 2.2204
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 267.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2677,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43",
            "2026-W44"
          ],
          "tokens_per_sec_best": 13.98,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 2.653,
          "first_at": "2026-10-24T22:41:42.807Z",
          "last_at": "2026-10-27T03:48:05.777Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.mr",
            "lang.nl",
            "lang.ru",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 2.2204,
          "usd_per_m_energy": 0.0737,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1774.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2688,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 614.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0505,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 248.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4506,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.0609,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.1053,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 901.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.285,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/ggml-org/gemma-3-1b-it-GGUF:Q8_0",
      "family": "gemma-3",
      "member": "1b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-3-1b-it",
      "hub_repo": "google/gemma-3-1b-it",
      "publisher": "ggml-org",
      "derivative": false,
      "cohort": 2,
      "rung": 1,
      "disk_mib": 1019,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-1b-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 3.2787
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 201.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3563,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 9.53,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 2.993,
          "first_at": "2026-10-26T03:00:53.560Z",
          "last_at": "2026-10-26T03:00:53.560Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.mr",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 3.2787,
          "usd_per_m_energy": 0.0831,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1337.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3568,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 462.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0675,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 187.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.599,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.3858,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.4643,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 679.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3785,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/google/gemma-4-12B-it-qat-q4_0-gguf:Q4_0",
      "family": "gemma-4",
      "member": "12b",
      "quant": "Q4_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-12B-it-qat-q4_0-unquantized",
      "hub_repo": "google/gemma-4-12B-it-qat-q4_0-unquantized",
      "publisher": "google",
      "derivative": false,
      "cohort": 2,
      "rung": 3,
      "disk_mib": 6652,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-12B-it-qat-q4_0-unquantized",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 1.2916
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 30.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.4744,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.0323,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 204.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.3519,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 27.74,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 5.824,
          "first_at": "2026-10-23T16:00:50.658Z",
          "last_at": "2026-10-23T16:00:50.658Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.2916,
          "usd_per_m_energy": 0.1618,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 28.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.0704,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 65.2869,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 38.0028,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 104.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.5145,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/hugging-quants/Llama-3.2-1B-Instruct-Q4_K_M-GGUF:Q4_K_M",
      "family": "llama-3-2",
      "member": "1b~hugging-quants-llama-3-2-k-m",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-1B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-1B-Instruct",
      "publisher": "hugging-quants",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 770,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-1B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-1b-instruct",
        "ids": [
          "meta-llama/llama-3.2-1b-instruct"
        ],
        "usd_per_m_in": 0.027,
        "usd_per_m_out": 0.201,
        "usd_per_m_cached_in": null,
        "context_length": 60000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 2.7594,
        "vs_api": 13.73
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 266.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2684,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 11.25,
          "effective_ctx_best": null,
          "joules_per_token_best": 3.738,
          "first_at": "2026-10-25T22:34:51.037Z",
          "last_at": "2026-10-25T22:34:51.037Z",
          "caps": [
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 2.7594,
          "usd_per_m_energy": 0.1038,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1770.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2695,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 612.59,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0507,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 248.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4518,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.075,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.1172,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 898.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2857,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/hugging-quants/Llama-3.2-3B-Instruct-Q4_K_M-GGUF:Q4_K_M",
      "family": "llama-3-2",
      "member": "3b~hugging-quants-llama-3-2-k-m",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-3B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-3B-Instruct",
      "publisher": "hugging-quants",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 1925,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 106.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6799,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 24.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2996,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 708.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6752,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 245.03,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1305,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 99.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1389,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.05,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 17.9108,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.4222,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 359.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.717,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/huihui-ai/Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF:Q2_K",
      "family": "qwen-3-6",
      "member": "35b-a3b~huihui-ai-abliterated-mtp",
      "quant": "Q2_K",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.6-35B-A3B",
      "hub_repo": "Qwen/Qwen3.6-35B-A3B",
      "publisher": "huihui-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 5,
      "disk_mib": 12632,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.6-35B-A3B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.6-35b-a3b",
        "ids": [
          "qwen/qwen3.6-35b-a3b"
        ],
        "usd_per_m_in": 0.1,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.025,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 8.1393,
        "vs_api": 9.04
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 189.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4284,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 5,
          "effective_ctx_best": null,
          "joules_per_token_best": 6.124,
          "first_at": "2026-11-05T16:20:03.598Z",
          "last_at": "2026-11-05T16:20:03.598Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.ta",
            "output.long",
            "reason.channel",
            "research.synthesize",
            "thinking",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 8.1393,
          "usd_per_m_energy": 0.1701,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1258.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3867,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 435.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0934,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 176.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.69,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.285,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.37,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.5685,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 639.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4169,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/huihui-ai/Huihui-Qwen3.6-35B-A3B-abliterated-MTP-GGUF:Q6_K",
      "family": "qwen-3-6",
      "member": "35b-a3b~huihui-ai-abliterated-mtp",
      "quant": "Q6_K",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.6-35B-A3B",
      "hub_repo": "Qwen/Qwen3.6-35B-A3B",
      "publisher": "huihui-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 27854,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.6-35B-A3B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.6-35b-a3b",
        "ids": [
          "qwen/qwen3.6-35b-a3b"
        ],
        "usd_per_m_in": 0.1,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.025,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 12.374,
        "vs_api": 13.75
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 86.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0886,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W45"
          ],
          "tokens_per_sec_best": 4.29,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 8.986,
          "first_at": "2026-10-28T06:08:09.160Z",
          "last_at": "2026-11-05T14:31:37.637Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.am",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.es",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ne",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.si",
            "lang.ta",
            "lang.te",
            "lang.ur",
            "lang.yo",
            "output.long",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 12.374,
          "usd_per_m_energy": 0.2496,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 570.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8743,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 197.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2687,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 80.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6762,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 28.7239,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 16.6916,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 289.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.962,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF:Q4_K_M",
      "family": "qwen-3-8",
      "member": "27b~huihui-ai-abliterated",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "huihui-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 15784,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 4.6269,
        "vs_api": 2.47
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.4425,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.4206,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 86.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.6667,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 9.35,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 22.428,
          "first_at": "2026-10-28T10:49:23.756Z",
          "last_at": "2026-10-28T10:49:23.756Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 4.6269,
          "usd_per_m_energy": 0.623,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.2703,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 171.0557,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 98.6086,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.1367,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/huihui-ai/Huihui-Qwen3.8-27B-abliterated-GGUF:Q4_K_S",
      "family": "qwen-3-8",
      "member": "27b~huihui-ai-abliterated",
      "quant": "Q4_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "huihui-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 14900,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 3.5931,
        "vs_api": 1.92
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.0303,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 13.3781,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 91.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.3411,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 11.84,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 24.323,
          "first_at": "2026-10-28T08:17:18.415Z",
          "last_at": "2026-10-28T08:17:18.415Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 3.5931,
          "usd_per_m_energy": 0.6756,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.6378,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 159.8036,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 92.331,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 46.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.7765,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:BF16",
      "family": "granite-4-0",
      "member": "1b",
      "quant": "BF16",
      "engine": "ollama",
      "modality": "text",
      "base": "ibm-granite/granite-4.0-1b",
      "hub_repo": "ibm-granite/granite-4.0-1b",
      "publisher": "ibm-granite",
      "derivative": false,
      "cohort": 1,
      "rung": 4,
      "disk_mib": 3116,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ibm-granite/granite-4.0-1b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 6.9963
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 65.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1153,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 4.71,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 7.84,
          "first_at": "2026-10-27T04:18:29.518Z",
          "last_at": "2026-10-27T04:18:29.518Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.ml",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call"
          ],
          "usd_per_m_out": 6.9963,
          "usd_per_m_energy": 0.2178,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 437.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0951,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 151.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2177,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 61.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8595,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 29.4032,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 17.0949,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 222.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.165,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q2_K",
      "family": "granite-4-0",
      "member": "1b",
      "quant": "Q2_K",
      "engine": "ollama",
      "modality": "text",
      "base": "ibm-granite/granite-4.0-1b",
      "hub_repo": "ibm-granite/granite-4.0-1b",
      "publisher": "ibm-granite",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 669,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ibm-granite/granite-4.0-1b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 3.9592
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 306.99,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.233,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 7.82,
          "effective_ctx_best": null,
          "joules_per_token_best": 5.851,
          "first_at": "2026-10-26T03:18:07.403Z",
          "last_at": "2026-10-26T03:18:07.403Z",
          "caps": [
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 3.9592,
          "usd_per_m_energy": 0.1625,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2037.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2341,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 705.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0439,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 285.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3923,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.144,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 20.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.5723,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1034.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2482,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q3_K_L",
      "family": "granite-4-0",
      "member": "1b",
      "quant": "Q3_K_L",
      "engine": "ollama",
      "modality": "text",
      "base": "ibm-granite/granite-4.0-1b",
      "hub_repo": "ibm-granite/granite-4.0-1b",
      "publisher": "ibm-granite",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 883,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ibm-granite/granite-4.0-1b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 4.3064
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 232.59,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3082,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 7.23,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 4.247,
          "first_at": "2026-10-26T03:53:03.477Z",
          "last_at": "2026-10-26T03:53:03.477Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 4.3064,
          "usd_per_m_energy": 0.118,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1543.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3091,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 534.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0583,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 216.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5185,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.1282,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.7258,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 783.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3278,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q3_K_S",
      "family": "granite-4-0",
      "member": "1b",
      "quant": "Q3_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "ibm-granite/granite-4.0-1b",
      "hub_repo": "ibm-granite/granite-4.0-1b",
      "publisher": "ibm-granite",
      "derivative": false,
      "cohort": 1,
      "rung": 2,
      "disk_mib": 749,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ibm-granite/granite-4.0-1b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 3.888
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 274.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2611,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 7.98,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 13.139,
          "first_at": "2026-10-26T03:33:20.172Z",
          "last_at": "2026-10-26T03:33:20.172Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call"
          ],
          "usd_per_m_out": 3.888,
          "usd_per_m_energy": 0.365,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1819.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2622,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 629.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0493,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 255.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4394,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.4,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.8829,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.0035,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 924.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2779,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q4_0",
      "family": "granite-4-0",
      "member": "1b",
      "quant": "Q4_0",
      "engine": "ollama",
      "modality": "text",
      "base": "ibm-granite/granite-4.0-1b",
      "hub_repo": "ibm-granite/granite-4.0-1b",
      "publisher": "ibm-granite",
      "derivative": false,
      "cohort": 1,
      "rung": 3,
      "disk_mib": 929,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ibm-granite/granite-4.0-1b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 3.5626
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 221.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3245,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 8.75,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 6.077,
          "first_at": "2026-10-26T04:12:49.423Z",
          "last_at": "2026-10-26T04:12:49.423Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 3.5626,
          "usd_per_m_energy": 0.1688,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1467.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3253,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 507.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0614,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 205.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5457,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.5493,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 14.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.9742,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 745.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3449,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/ibm-granite/granite-4.0-1b-GGUF:Q4_K_M",
      "family": "granite-4-0",
      "member": "1b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "ibm-granite/granite-4.0-1b",
      "hub_repo": "ibm-granite/granite-4.0-1b",
      "publisher": "ibm-granite",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 976,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ibm-granite/granite-4.0-1b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 2.745
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 210.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3411,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43",
            "2026-W44"
          ],
          "tokens_per_sec_best": 11.37,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 3.814,
          "first_at": "2026-10-24T23:11:38.721Z",
          "last_at": "2026-10-27T01:03:58.424Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.nl",
            "lang.ru",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 2.745,
          "usd_per_m_energy": 0.1059,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1396.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3418,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 483.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0646,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 195.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5735,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.9933,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.2308,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 709.22,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3624,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/ibm-granite/granite-4.0-1b-speech-GGUF:Q4_K_M",
      "family": "granite-4-0",
      "member": "speech-1b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "ibm-granite/granite-4.0-1b-speech",
      "hub_repo": "ibm-granite/granite-4.0-1b-speech",
      "publisher": "ibm-granite",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 1086,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ibm-granite/granite-4.0-1b-speech",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.2413
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 189.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.38,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7176,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1255.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3803,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 129.7,
          "effective_ctx_best": null,
          "joules_per_token_best": 3.87,
          "first_at": "2026-11-02T22:09:43.714Z",
          "last_at": "2026-11-02T22:09:43.714Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.2413,
          "usd_per_m_energy": 0.1075,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 176.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6387,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.0218,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.8277,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 637.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4034,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/ibm-granite/granite-4.2-3b-GGUF:Q4_K_M",
      "family": "granite-4-2",
      "member": "3b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "ibm-granite/granite-4.2-3b",
      "hub_repo": "ibm-granite/granite-4.2-3b",
      "publisher": "ibm-granite",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 2140,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ibm-granite/granite-4.2-3b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 6.7142
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 95.97,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7577,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4525,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 636.99,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7508,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 220.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1459,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 16.87,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 19.529,
          "first_at": "2026-11-06T14:45:19.833Z",
          "last_at": "2026-11-06T14:45:19.833Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.kn",
            "lang.ta",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 6.7142,
          "usd_per_m_energy": 0.5425,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19.9763,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.6156,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 323.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7976,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/ibm-granite/granite-4.2-8b-GGUF:Q4_K_M",
      "family": "granite-4-2",
      "member": "8b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "ibm-granite/granite-4.2-8b",
      "hub_repo": "ibm-granite/granite-4.2-8b",
      "publisher": "ibm-granite",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 5100,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ibm-granite/granite-4.2-8b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "ibm-granite/granite-4.2-8b",
        "ids": [
          "ibm-granite/granite-4.2-8b"
        ],
        "usd_per_m_in": 0.06,
        "usd_per_m_out": 0.15,
        "usd_per_m_cached_in": 0.015,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 40.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8655,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.7209,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 267.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7983,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 92.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3737,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 37.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.0864,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 49.0997,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.63,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 28.5637,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 135.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.9187,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q3_K_M",
      "family": "qwen-3-6",
      "member": "35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved",
      "quant": "Q3_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
      "hub_repo": "llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
      "publisher": "llmfan46",
      "derivative": true,
      "cohort": 3,
      "rung": 2,
      "disk_mib": 16467,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
        "fetched_at": "2026-11-09T10:51:05.097Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 9.072
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 145.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5799,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 4.83,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 7.659,
          "first_at": "2026-11-05T10:40:30.734Z",
          "last_at": "2026-11-05T10:40:30.734Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 9.072,
          "usd_per_m_energy": 0.2127,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 965.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5073,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 334.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1311,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 135.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9225,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 15.285,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.8907,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 490.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5498,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q4_K_M",
      "family": "qwen-3-6",
      "member": "35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
      "hub_repo": "llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
      "publisher": "llmfan46",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 20771,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
        "fetched_at": "2026-11-09T10:51:05.097Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 9.7567
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 115.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7618,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 4.85,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 12.472,
          "first_at": "2026-11-05T09:14:04.937Z",
          "last_at": "2026-11-05T09:14:04.937Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 9.7567,
          "usd_per_m_energy": 0.3464,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 765.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6445,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 264.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1786,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 107.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1963,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 20.0629,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.6856,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 388.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7026,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved-GGUF:Q4_K_S",
      "family": "qwen-3-6",
      "member": "35b-a3b~llmfan46-uncensored-heretic-native-mtp-preserved",
      "quant": "Q4_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
      "hub_repo": "llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
      "publisher": "llmfan46",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 19487,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/llmfan46/Qwen3.6-35B-A3B-uncensored-heretic-Native-MTP-Preserved",
        "fetched_at": "2026-11-09T10:51:05.097Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 9.6206
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 122.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7062,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 4.81,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 9.751,
          "first_at": "2026-11-05T12:20:58.138Z",
          "last_at": "2026-11-05T12:20:58.138Z",
          "caps": [
            "code.reason",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 9.6206,
          "usd_per_m_energy": 0.2709,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 816.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6033,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 282.4,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1639,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 114.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1131,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 18.6332,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.8268,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 414.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6566,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/lm-kit/embeddinggemma-300m-gguf:F16",
      "family": "embeddinggemma",
      "member": "300m~lm-kit-embeddinggemma",
      "quant": "F16",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "lm-kit/embeddinggemma-300m-gguf",
      "publisher": "lm-kit",
      "derivative": true,
      "cohort": 3,
      "rung": 4,
      "disk_mib": 584,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/lm-kit/embeddinggemma-300m-gguf",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 351.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2032,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:37:13.045Z",
          "last_at": "2026-10-25T20:37:13.045Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 81.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3809,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2334.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2044,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 807.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0382,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 327.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3422,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.3559,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.1159,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1185.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2166,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/lm-kit/embeddinggemma-300m-gguf:Q4_K_M",
      "family": "embeddinggemma",
      "member": "300m~lm-kit-embeddinggemma",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "lm-kit/embeddinggemma-300m-gguf",
      "publisher": "lm-kit",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 225,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/lm-kit/embeddinggemma-300m-gguf",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 912.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.078,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:34:26.819Z",
          "last_at": "2026-10-25T20:34:26.819Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 210.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1453,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6058.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0787,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2096.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0146,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 849.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1315,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 34.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.0553,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 59.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1953,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3076.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0834,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/lm-kit/embeddinggemma-300m-gguf:Q5_K_M",
      "family": "embeddinggemma",
      "member": "300m~lm-kit-embeddinggemma",
      "quant": "Q5_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "lm-kit/embeddinggemma-300m-gguf",
      "publisher": "lm-kit",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 235,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/lm-kit/embeddinggemma-300m-gguf",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 873.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0814,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:35:14.487Z",
          "last_at": "2026-10-25T20:35:14.487Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 201.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1518,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5800.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0822,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2007.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0152,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 813.37,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1374,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 33.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.1467,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 56.99,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2487,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2945.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0871,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/lm-kit/embeddinggemma-300m-gguf:Q8_0",
      "family": "embeddinggemma",
      "member": "300m~lm-kit-embeddinggemma",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "lm-kit/embeddinggemma-300m-gguf",
      "publisher": "lm-kit",
      "derivative": true,
      "cohort": 3,
      "rung": 1,
      "disk_mib": 313,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/lm-kit/embeddinggemma-300m-gguf",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 656.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1086,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:36:09.280Z",
          "last_at": "2026-10-25T20:36:09.280Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 151.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2027,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4355.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1095,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1507,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0204,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 610.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.183,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 24.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.8617,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 42.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6646,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2211.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.116,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q3_K_L",
      "family": "phi-4",
      "member": "mini-reasoning",
      "quant": "Q3_K_L",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "lmstudio-community/Phi-4-mini-reasoning-GGUF",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W37",
        "2026-W38",
        "2026-W41"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W41"
          ],
          "tokens_per_sec_best": 3.42,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-08T13:24:57.457Z",
          "last_at": "2026-10-08T13:24:57.457Z",
          "caps": [
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking"
          ],
          "verdict": "only the network runs it"
        },
        "windows/amd64/amd/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 3.85,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T08:57:05.460Z",
          "last_at": "2026-09-15T08:57:05.460Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 89.92,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-13T18:23:49.162Z",
          "last_at": "2026-09-13T18:23:49.162Z",
          "caps": [
            "code.reason",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q4_K_M",
      "family": "phi-4",
      "member": "mini-reasoning",
      "quant": "Q4_K_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "lmstudio-community/Phi-4-mini-reasoning-GGUF",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W38",
        "2026-W41"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W41"
          ],
          "tokens_per_sec_best": 7.26,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-08T14:06:13.742Z",
          "last_at": "2026-10-08T14:06:13.742Z",
          "caps": [
            "code.reason",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking"
          ],
          "verdict": "only the network runs it"
        },
        "windows/amd64/amd/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 4.75,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T09:38:46.908Z",
          "last_at": "2026-09-15T09:38:46.908Z",
          "caps": [
            "code.reason",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q6_K",
      "family": "phi-4",
      "member": "mini-reasoning",
      "quant": "Q6_K",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "lmstudio-community/Phi-4-mini-reasoning-GGUF",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/amd64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-28T01:42:54.332Z",
          "last_at": "2026-08-28T01:42:54.332Z",
          "caps": [],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/lmstudio-community/Phi-4-mini-reasoning-GGUF:Q8_0",
      "family": "phi-4",
      "member": "mini-reasoning",
      "quant": "Q8_0",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "lmstudio-community/Phi-4-mini-reasoning-GGUF",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W40"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W40"
          ],
          "tokens_per_sec_best": 22.03,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-09-29T10:59:33.267Z",
          "last_at": "2026-09-29T10:59:33.267Z",
          "caps": [
            "code.reason",
            "lang.en",
            "lang.ta",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/lmstudio-community/Phi-4-reasoning-plus-GGUF:Q4_K_M",
      "family": "phi-4",
      "member": "reasoning-plus",
      "quant": "Q4_K_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "lmstudio-community/Phi-4-reasoning-plus-GGUF",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W38"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T04:44:02.843Z",
          "last_at": "2026-09-15T04:44:02.843Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/lmstudio-community/Phi-4-reasoning-plus-GGUF:Q6_K",
      "family": "phi-4",
      "member": "reasoning-plus",
      "quant": "Q6_K",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "lmstudio-community/Phi-4-reasoning-plus-GGUF",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W38"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 3.26,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-16T09:54:24.719Z",
          "last_at": "2026-09-16T09:54:24.719Z",
          "caps": [
            "code.reason",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/lmstudio-community/gemma-3-1b-it-GGUF:Q3_K_L",
      "family": "gemma-3",
      "member": "1b",
      "quant": "Q3_K_L",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-3-1b-it",
      "hub_repo": "google/gemma-3-1b-it",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": 2,
      "rung": 9,
      "disk_mib": 716,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-1b-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 2.3664
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 286.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2495,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 13.1,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 2.266,
          "first_at": "2026-10-25T23:25:24.142Z",
          "last_at": "2026-10-25T23:25:24.142Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 2.3664,
          "usd_per_m_energy": 0.0629,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1903.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2506,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 658.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0471,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 266.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.42,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.5767,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 18.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.8244,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 966.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2657,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/lmstudio-community/gemma-3-1b-it-GGUF:Q6_K",
      "family": "gemma-3",
      "member": "1b",
      "quant": "Q6_K",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-3-1b-it",
      "hub_repo": "google/gemma-3-1b-it",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": 2,
      "rung": 3,
      "disk_mib": 964,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-1b-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 2.8917
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 213.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3368,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 10.79,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 4.164,
          "first_at": "2026-10-25T23:47:15.858Z",
          "last_at": "2026-10-25T23:47:15.858Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.mr",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 2.8917,
          "usd_per_m_energy": 0.1157,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1414.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3375,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 489.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0638,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 198.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5664,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.8808,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.1661,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 718.05,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.358,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/meta-models/Muse-Glimmer-30B-GGUF:Q4_K_M",
      "family": "muse-glimmer",
      "member": "30b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-models/Muse-Glimmer-30B",
      "hub_repo": "meta-models/Muse-Glimmer-30B",
      "publisher": "meta-models",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 15980,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/meta-models/Muse-Glimmer-30B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "meta/muse-glimmer-30b",
        "ids": [
          "meta/muse-glimmer-30b",
          "meta/muse-glimmer-30b:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.04,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.96,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 18.0482,
        "vs_api": 15.04
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.5352,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.6693,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 85.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.7389,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 29.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4709,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 6.9,
          "effective_ctx_best": null,
          "joules_per_token_best": 333.409,
          "first_at": "2026-11-06T15:00:58.056Z",
          "last_at": "2026-11-06T15:00:58.056Z",
          "caps": [],
          "usd_per_m_out": 18.0482,
          "usd_per_m_energy": 9.2614,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 171.3812,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 99.9724,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.32,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.2155,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:F16",
      "family": "functiongemma",
      "member": "270m~mingxilei-unsloth-ft-qat",
      "quant": "F16",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
      "publisher": "mingxilei",
      "derivative": true,
      "cohort": 3,
      "rung": 4,
      "disk_mib": 517,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 1.3332
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 397.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1797,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 23.13,
          "effective_ctx_best": null,
          "joules_per_token_best": 3.676,
          "first_at": "2026-10-25T04:28:14.495Z",
          "last_at": "2026-10-25T04:28:14.495Z",
          "caps": [
            "code.reason",
            "code.write",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.3332,
          "usd_per_m_energy": 0.1021,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2636.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1809,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 912.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0338,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 369.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3028,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.7374,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 25.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7554,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1338.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1917,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q4_K_M",
      "family": "functiongemma",
      "member": "270m~mingxilei-unsloth-ft-qat",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
      "publisher": "mingxilei",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 241,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.779
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 852.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0835,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 39.3,
          "effective_ctx_best": null,
          "joules_per_token_best": 5.353,
          "first_at": "2026-10-25T03:03:15.910Z",
          "last_at": "2026-10-25T03:03:15.910Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.779,
          "usd_per_m_energy": 0.1487,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5656.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0843,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1957.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0156,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 793.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1409,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 32.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.2013,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 55.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2805,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2872.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0893,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q5_K_M",
      "family": "functiongemma",
      "member": "270m~mingxilei-unsloth-ft-qat",
      "quant": "Q5_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
      "publisher": "mingxilei",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 247,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9625
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 831.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0856,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 31.81,
          "effective_ctx_best": null,
          "joules_per_token_best": 3.369,
          "first_at": "2026-10-25T03:23:23.866Z",
          "last_at": "2026-10-25T03:23:23.866Z",
          "caps": [
            "code.write",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.9625,
          "usd_per_m_energy": 0.0936,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5518.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0864,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1909.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.016,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 773.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1444,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 31.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.2566,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 54.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.3124,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2802.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0915,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF:Q8_0",
      "family": "functiongemma",
      "member": "270m~mingxilei-unsloth-ft-qat",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": null,
      "hub_repo": "mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
      "publisher": "mingxilei",
      "derivative": true,
      "cohort": 3,
      "rung": 1,
      "disk_mib": 278,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mingxilei/functiongemma-270m-it-unsloth-bnb-4bit-ft-QAT-GGUF",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9734
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 738.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0964,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 31.48,
          "effective_ctx_best": null,
          "joules_per_token_best": 0.854,
          "first_at": "2026-10-25T03:39:40.913Z",
          "last_at": "2026-10-25T03:39:40.913Z",
          "caps": [
            "code.write",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.9734,
          "usd_per_m_energy": 0.0237,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4903.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0972,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1696.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0181,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 687.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1625,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 28.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.541,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 48.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4778,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2489.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.103,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mistralai/Ministral-3-14B-Instruct-2512-GGUF:Q4_K_M",
      "family": "ministral-3",
      "member": "14b-instruct-2512",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "mistralai/Ministral-3-14B-Instruct-2512",
      "hub_repo": "mistralai/Ministral-3-14B-Instruct-2512",
      "publisher": "mistralai",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 7857,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Ministral-3-14B-Instruct-2512",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "mistralai/ministral-14b-2512",
        "ids": [
          "mistralai/ministral-14b-2512"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 0.2,
        "usd_per_m_cached_in": 0.02,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 24.0593,
        "vs_api": 120.3
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 26.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.9597,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 1.53,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 27.097,
          "first_at": "2026-10-30T01:50:29.183Z",
          "last_at": "2026-10-30T01:50:29.183Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 24.0593,
          "usd_per_m_energy": 0.7527,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 173.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7834,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 60.03,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6132,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 24.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.8468,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.99,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 78.1477,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 45.5096,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 88.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.9812,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mistralai/Ministral-3-3B-Instruct-2512-GGUF:Q4_K_M",
      "family": "ministral-3",
      "member": "3b-instruct-2512",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "mistralai/Ministral-3-3B-Instruct-2512",
      "hub_repo": "mistralai/Ministral-3-3B-Instruct-2512",
      "publisher": "mistralai",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 2047,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Ministral-3-3B-Instruct-2512",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "mistralai/ministral-3b-2512",
        "ids": [
          "mistralai/ministral-3b-2512"
        ],
        "usd_per_m_in": 0.1,
        "usd_per_m_out": 0.1,
        "usd_per_m_cached_in": 0.01,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 6.8561,
        "vs_api": 68.56
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 100.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.724,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 23.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.3865,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 665.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7181,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 230.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1392,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 16.51,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 27.755,
          "first_at": "2026-11-06T12:05:00.442Z",
          "last_at": "2026-11-06T12:05:00.442Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.te",
            "output.long",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 6.8561,
          "usd_per_m_energy": 0.771,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.81,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19.0651,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 4.66,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 34.398,
          "first_at": "2026-11-08T11:18:30.720Z",
          "last_at": "2026-11-08T11:18:30.720Z",
          "caps": [
            "tools",
            "tools.call"
          ],
          "usd_per_m_out": 15.5876,
          "usd_per_m_energy": 0.9555,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 338.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7627,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mistralai/Ministral-3-8B-Instruct-2512-GGUF:Q4_K_M",
      "family": "ministral-3",
      "member": "8b-instruct-2512",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "mistralai/Ministral-3-8B-Instruct-2512",
      "hub_repo": "mistralai/Ministral-3-8B-Instruct-2512",
      "publisher": "mistralai",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 4958,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Ministral-3-8B-Instruct-2512",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/ministral-8b-2512",
        "ids": [
          "mistralai/ministral-8b-2512"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.15,
        "usd_per_m_cached_in": 0.015,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 41.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8109,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.6075,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 274.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7479,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 95.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3621,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 38.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.9977,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 47.7752,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 27.7804,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 139.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8644,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Airavata-GGUF:Q4_K_M",
      "family": "airavata",
      "member": "7b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "ai4bharat/Airavata",
      "hub_repo": "ai4bharat/Airavata",
      "publisher": "mradermacher",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 3979,
      "active_share": 1,
      "licence": {
        "id": "llama2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/ai4bharat/Airavata",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 1.0654
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 51.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4379,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.8258,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 342.59,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4004,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 31.59,
          "effective_ctx_best": null,
          "joules_per_token_best": 3.958,
          "first_at": "2026-10-20T08:11:09.520Z",
          "last_at": "2026-10-20T08:11:09.520Z",
          "caps": [
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long"
          ],
          "usd_per_m_out": 1.0654,
          "usd_per_m_energy": 0.1099,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 52.6,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 2.385,
          "first_at": "2026-11-03T21:33:27.602Z",
          "last_at": "2026-11-03T21:33:27.602Z",
          "caps": [
            "code.reason",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long"
          ],
          "usd_per_m_out": 2.1819,
          "usd_per_m_energy": 0.0663,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 37.8624,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.37,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 22.0209,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 173.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4917,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/GLM-4.5V-i1-GGUF:Q4_K_M",
      "family": "glm-4-5",
      "member": "v~mradermacher-i1",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "zai-org/GLM-4.5V",
      "hub_repo": "zai-org/GLM-4.5V",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 67184,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.5V",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.5v",
        "ids": [
          "z-ai/glm-4.5v"
        ],
        "usd_per_m_in": 0.6,
        "usd_per_m_out": 1.8,
        "usd_per_m_cached_in": 0.11,
        "context_length": 65536,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 41.061,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 119.8467,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 20.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.1805,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.1212,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 58.3166,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1047.0558,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 628.2335,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 30.1868,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/GLM-4.6V-i1-GGUF:Q4_K_M",
      "family": "glm-4-6",
      "member": "v~mradermacher-i1",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "zai-org/GLM-4.6V",
      "hub_repo": "zai-org/GLM-4.6V",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 67184,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.6V",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.6v",
        "ids": [
          "z-ai/glm-4.6v"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.055,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 41.061,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 119.8467,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 20.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.1805,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.1212,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 58.3166,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1047.0558,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 628.2335,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 30.1868,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q2_K",
      "family": "glm-ocr",
      "member": "base",
      "quant": "Q2_K",
      "engine": "ollama",
      "modality": "text",
      "base": "zai-org/GLM-OCR",
      "hub_repo": "zai-org/GLM-OCR",
      "publisher": "mradermacher",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 344,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-OCR",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "windows/amd64/intel/16g",
        "usd_per_m_out": 5.4143
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 597.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1193,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 137.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2229,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3962.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1203,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1371.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0224,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 555.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2012,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.1458,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": 13.16,
          "effective_ctx_best": null,
          "joules_per_token_best": 75.114,
          "first_at": "2026-11-14T01:45:47.197Z",
          "last_at": "2026-11-14T01:45:47.197Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output",
            "vision",
            "vision.multi_image",
            "vision.read"
          ],
          "usd_per_m_out": 5.4143,
          "usd_per_m_energy": 2.0865,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2012.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1275,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q3_K_S",
      "family": "glm-ocr",
      "member": "base",
      "quant": "Q3_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "zai-org/GLM-OCR",
      "hub_repo": "zai-org/GLM-OCR",
      "publisher": "mradermacher",
      "derivative": false,
      "cohort": 1,
      "rung": 2,
      "disk_mib": 402,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-OCR",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "windows/amd64/intel/16g",
        "usd_per_m_out": 3.937
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 510.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1396,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 117.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2609,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3390.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1406,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1173.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0262,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 475.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2352,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.679,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": 18.11,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 99.633,
          "first_at": "2026-11-14T17:42:24.397Z",
          "last_at": "2026-11-14T17:42:24.397Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output",
            "vision",
            "vision.multi_image",
            "vision.read"
          ],
          "usd_per_m_out": 3.937,
          "usd_per_m_energy": 2.7676,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1721.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.149,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/GLM-OCR-GGUF:Q4_K_M",
      "family": "glm-ocr",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "zai-org/GLM-OCR",
      "hub_repo": "zai-org/GLM-OCR",
      "publisher": "mradermacher",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 523,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-OCR",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 1.0552
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 392.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1818,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 90.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3405,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2606.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.183,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 901.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0342,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 106.1,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 19.162,
          "first_at": "2026-11-06T13:35:55.849Z",
          "last_at": "2026-11-06T13:35:55.849Z",
          "caps": [
            "code.reason",
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output",
            "vision",
            "vision.multi_image",
            "vision.read"
          ],
          "usd_per_m_out": 1.0552,
          "usd_per_m_energy": 0.5323,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 14.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.7918,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": 12.98,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 55.445,
          "first_at": "2026-11-10T20:26:21.323Z",
          "last_at": "2026-11-10T20:26:21.323Z",
          "caps": [
            "code.reason",
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output",
            "vision",
            "vision.multi_image",
            "vision.read"
          ],
          "usd_per_m_out": 5.5006,
          "usd_per_m_energy": 1.5401,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1323.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1939,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/HY-MT1.5-1.8B-i1-GGUF:Q4_0",
      "family": "hy-mt-1-5",
      "member": "1.8b~mradermacher-i1",
      "quant": "Q4_0",
      "engine": "ollama",
      "modality": "text",
      "base": "tencent/HY-MT1.5-1.8B",
      "hub_repo": "tencent/HY-MT1.5-1.8B",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 1029,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/HY-MT1.5-1.8B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.2367
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 199.59,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3598,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 46.03,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.679,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1324.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3603,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 132.06,
          "effective_ctx_best": null,
          "joules_per_token_best": 207.768,
          "first_at": "2026-10-28T04:48:47.080Z",
          "last_at": "2026-10-28T04:48:47.080Z",
          "caps": [],
          "usd_per_m_out": 0.2367,
          "usd_per_m_energy": 5.7713,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 48.53,
          "effective_ctx_best": null,
          "joules_per_token_best": 257,
          "first_at": "2026-11-03T21:20:41.123Z",
          "last_at": "2026-11-03T21:20:41.123Z",
          "caps": [],
          "usd_per_m_out": 2.3154,
          "usd_per_m_energy": 7.1389,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.4861,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.5153,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 672.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3822,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:F16",
      "family": "functiongemma",
      "member": "270m~acon96-home",
      "quant": "F16",
      "engine": "ollama",
      "modality": "text",
      "base": "acon96/Home-FunctionGemma-270m",
      "hub_repo": "acon96/Home-FunctionGemma-270m",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 4,
      "disk_mib": 517,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/acon96/Home-FunctionGemma-270m",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 1.3681
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 397.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1797,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 22.54,
          "effective_ctx_best": null,
          "joules_per_token_best": 5.033,
          "first_at": "2026-10-25T04:17:23.155Z",
          "last_at": "2026-10-25T04:17:23.155Z",
          "caps": [],
          "usd_per_m_out": 1.3681,
          "usd_per_m_energy": 0.1398,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2636.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1809,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 912.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0338,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 369.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3028,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.7374,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 25.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7554,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1338.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1917,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:IQ4_XS",
      "family": "functiongemma",
      "member": "270m~acon96-home",
      "quant": "IQ4_XS",
      "engine": "ollama",
      "modality": "text",
      "base": "acon96/Home-FunctionGemma-270m",
      "hub_repo": "acon96/Home-FunctionGemma-270m",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 9,
      "disk_mib": 230,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/acon96/Home-FunctionGemma-270m",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.8792
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 892.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0797,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 34.81,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.72,
          "first_at": "2026-10-25T02:39:27.290Z",
          "last_at": "2026-10-25T02:39:27.290Z",
          "caps": [],
          "usd_per_m_out": 0.8792,
          "usd_per_m_energy": 0.0756,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5926.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0804,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2050.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0149,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 831.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1344,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 33.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.101,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 58.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.222,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3009.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0852,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q2_K",
      "family": "functiongemma",
      "member": "270m~acon96-home",
      "quant": "Q2_K",
      "engine": "ollama",
      "modality": "text",
      "base": "acon96/Home-FunctionGemma-270m",
      "hub_repo": "acon96/Home-FunctionGemma-270m",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 5,
      "disk_mib": 226,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/acon96/Home-FunctionGemma-270m",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.8865
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 908.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0783,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 34.52,
          "effective_ctx_best": null,
          "joules_per_token_best": 4.516,
          "first_at": "2026-10-25T02:29:13.928Z",
          "last_at": "2026-10-25T02:29:13.928Z",
          "caps": [],
          "usd_per_m_out": 0.8865,
          "usd_per_m_energy": 0.1254,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6031.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.079,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2087.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0147,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 845.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1321,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 34.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.0643,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 59.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2007,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3062.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0837,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q3_K_S",
      "family": "functiongemma",
      "member": "270m~acon96-home",
      "quant": "Q3_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "acon96/Home-FunctionGemma-270m",
      "hub_repo": "acon96/Home-FunctionGemma-270m",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 2,
      "disk_mib": 225,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/acon96/Home-FunctionGemma-270m",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.8528
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 912.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.078,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 35.88,
          "effective_ctx_best": null,
          "joules_per_token_best": 3.776,
          "first_at": "2026-10-25T02:16:10.838Z",
          "last_at": "2026-10-25T02:16:10.838Z",
          "caps": [],
          "usd_per_m_out": 0.8528,
          "usd_per_m_energy": 0.1049,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6058.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0787,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2096.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0146,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 849.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1315,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 34.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.0553,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 59.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1953,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3076.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0834,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q4_K_M",
      "family": "functiongemma",
      "member": "270m~acon96-home",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "acon96/Home-FunctionGemma-270m",
      "hub_repo": "acon96/Home-FunctionGemma-270m",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 241,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/acon96/Home-FunctionGemma-270m",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9199
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 852.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0835,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 33.28,
          "effective_ctx_best": null,
          "joules_per_token_best": 5.15,
          "first_at": "2026-10-25T03:02:13.593Z",
          "last_at": "2026-10-25T03:02:13.593Z",
          "caps": [],
          "usd_per_m_out": 0.9199,
          "usd_per_m_energy": 0.1431,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5656.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0843,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1957.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0156,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 793.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1409,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 32.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.2013,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 55.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2805,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2872.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0893,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q4_K_S",
      "family": "functiongemma",
      "member": "270m~acon96-home",
      "quant": "Q4_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "acon96/Home-FunctionGemma-270m",
      "hub_repo": "acon96/Home-FunctionGemma-270m",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 238,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/acon96/Home-FunctionGemma-270m",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9332
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 862.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0825,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 32.8,
          "effective_ctx_best": null,
          "joules_per_token_best": 4.382,
          "first_at": "2026-10-25T02:51:31.041Z",
          "last_at": "2026-10-25T02:51:31.041Z",
          "caps": [],
          "usd_per_m_out": 0.9332,
          "usd_per_m_energy": 0.1217,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5727.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0832,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1981.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0154,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 803.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1391,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 32.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.1743,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 56.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2645,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2908.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0882,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Home-FunctionGemma-270m-GGUF:Q8_0",
      "family": "functiongemma",
      "member": "270m~acon96-home",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "acon96/Home-FunctionGemma-270m",
      "hub_repo": "acon96/Home-FunctionGemma-270m",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 1,
      "disk_mib": 278,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/acon96/Home-FunctionGemma-270m",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9579
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 738.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0964,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 31.99,
          "effective_ctx_best": null,
          "joules_per_token_best": 5.164,
          "first_at": "2026-10-25T03:31:44.551Z",
          "last_at": "2026-10-25T03:31:44.551Z",
          "caps": [],
          "usd_per_m_out": 0.9579,
          "usd_per_m_energy": 0.1434,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4903.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0972,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1696.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0181,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 687.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1625,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 28.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.541,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 48.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4778,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2489.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.103,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Hy-MT2-1.8B-GGUF:Q3_K_S",
      "family": "hy-mt-2",
      "member": "1.8b",
      "quant": "Q3_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "tencent/Hy-MT2-1.8B",
      "hub_repo": "tencent/Hy-MT2-1.8B",
      "publisher": "mradermacher",
      "derivative": false,
      "cohort": 1,
      "rung": 2,
      "disk_mib": 831,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy-MT2-1.8B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [
        "2026-W46"
      ],
      "api": {
        "openrouter_id": "tencent/hy-mt2-1.8b",
        "ids": [
          "tencent/hy-mt2-1.8b"
        ],
        "usd_per_m_in": 0.044,
        "usd_per_m_out": 0.177,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "windows/amd64/intel/16g",
        "usd_per_m_out": 4.6798,
        "vs_api": 26.44
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 247.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2899,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5455,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1640.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2909,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 567.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0548,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 230.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4878,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.37,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.6466,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": 15.31,
          "effective_ctx_best": null,
          "joules_per_token_best": 5.077,
          "first_at": "2026-11-14T03:54:51.348Z",
          "last_at": "2026-11-14T03:54:51.348Z",
          "caps": [],
          "usd_per_m_out": 4.6798,
          "usd_per_m_energy": 0.141,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 832.97,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3084,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Hy-MT2-7B-i1-GGUF:Q4_K_M",
      "family": "hy-mt-2",
      "member": "7b~mradermacher-i1",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "tencent/Hy-MT2-7B",
      "hub_repo": "tencent/Hy-MT2-7B",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 4410,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy-MT2-7B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy-mt2-7b",
        "ids": [
          "tencent/hy-mt2-7b"
        ],
        "usd_per_m_in": 0.074,
        "usd_per_m_out": 0.295,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 46.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6011,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.1663,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 309.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.5532,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 106.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3179,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.6561,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 42.1249,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 24.5267,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 156.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6554,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Kimi-Linear-REAP-35B-A3B-Instruct-i1-GGUF:Q4_K_M",
      "family": "kimi-linear",
      "member": "35b-a3b~cerebras-reap",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "cerebras/Kimi-Linear-REAP-35B-A3B-Instruct",
      "hub_repo": "cerebras/Kimi-Linear-REAP-35B-A3B-Instruct",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 20277,
      "active_share": 0.0857,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/cerebras/Kimi-Linear-REAP-35B-A3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 9.8775
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 118.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7403,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 4.75,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 47.63,
          "first_at": "2026-10-30T15:45:09.494Z",
          "last_at": "2026-10-30T15:45:09.494Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 9.8775,
          "usd_per_m_energy": 1.3231,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 784.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6286,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 271.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1729,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 109.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1641,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19.5254,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.3455,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 398.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6848,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/MiniMax-M2.5-GGUF:Q4_K_M",
      "family": "minimax-m2-5",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.5",
      "hub_repo": "MiniMaxAI/MiniMax-M2.5",
      "publisher": "mradermacher",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 131933,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.5",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.5",
        "ids": [
          "minimax/minimax-m2.5"
        ],
        "usd_per_m_in": 0.27,
        "usd_per_m_out": 0.95,
        "usd_per_m_cached_in": 0.03,
        "context_length": 198000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 114.3202,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 382.7334,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 56.5242,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 38.4872,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 150.9622,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2972.3262,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1783.3957,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.25,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 69.2604,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/MiniMax-M2.5-i1-GGUF:Q4_K_M",
      "family": "minimax-m2-5",
      "member": "base~mradermacher-i1",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.5",
      "hub_repo": "MiniMaxAI/MiniMax-M2.5",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 131933,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.5",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.5",
        "ids": [
          "minimax/minimax-m2.5"
        ],
        "usd_per_m_in": 0.27,
        "usd_per_m_out": 0.95,
        "usd_per_m_cached_in": 0.03,
        "context_length": 198000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 114.3202,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 382.7334,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 56.5242,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 38.4872,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 150.9622,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2972.3262,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1783.3957,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.25,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 69.2604,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/MiniMax-M2.7-GGUF:Q4_K_M",
      "family": "minimax-m2-7",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.7",
      "hub_repo": "MiniMaxAI/MiniMax-M2.7",
      "publisher": "mradermacher",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 131933,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "other",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.7",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.7",
        "ids": [
          "minimax/minimax-m2.7"
        ],
        "usd_per_m_in": 0.21,
        "usd_per_m_out": 0.84,
        "usd_per_m_cached_in": 0.042,
        "context_length": 196608,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 114.3202,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 382.7334,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 56.5242,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 38.4872,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 150.9622,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2972.3262,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1783.3957,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.25,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 69.2604,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/MiniMax-M2.7-i1-GGUF:Q4_K_M",
      "family": "minimax-m2-7",
      "member": "base~mradermacher-i1",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.7",
      "hub_repo": "MiniMaxAI/MiniMax-M2.7",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 131933,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "other",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.7",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.7",
        "ids": [
          "minimax/minimax-m2.7"
        ],
        "usd_per_m_in": 0.21,
        "usd_per_m_out": 0.84,
        "usd_per_m_cached_in": 0.042,
        "context_length": 196608,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 114.3202,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 382.7334,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 56.5242,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 38.4872,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 150.9622,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2972.3262,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1783.3957,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.25,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 69.2604,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Mistral-Large-3-675B-Instruct-2512-GGUF:Q4_K_M",
      "family": "mistral-large-3",
      "member": "675b-instruct-2512",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "mistralai/Mistral-Large-3-675B-Instruct-2512",
      "hub_repo": "mistralai/Mistral-Large-3-675B-Instruct-2512",
      "publisher": "mradermacher",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 386958,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Large-3-675B-Instruct-2512",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-large-2512",
        "ids": [
          "mistralai/mistral-large-2512"
        ],
        "usd_per_m_in": 0.5,
        "usd_per_m_out": 1.5,
        "usd_per_m_cached_in": 0.05,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 728.0744,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2877.6991,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 224.8395,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.22,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 283.0524,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 870.2755,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19293.9724,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.03,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12862.6483,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 319.0822,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Mistral-Medium-3.5-128B-GGUF:Q4_K_M",
      "family": "mistral-medium-3-5",
      "member": "128b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "mistralai/Mistral-Medium-3.5-128B",
      "hub_repo": "mistralai/Mistral-Medium-3.5-128B",
      "publisher": "mradermacher",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 71427,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Medium-3.5-128B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-medium-3-5",
        "ids": [
          "mistralai/mistral-medium-3-5",
          "mistralai/mistral-medium-3-5:batch"
        ],
        "usd_per_m_in": 1.5,
        "usd_per_m_out": 7.5,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": 3.75,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 44.8263,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 134.1577,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 28.0218,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 13.4158,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 63.3042,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1173.6333,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 679.4719,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 32.4435,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Mistral-Medium-3.5-128B-i1-GGUF:Q4_K_M",
      "family": "mistral-medium-3-5",
      "member": "128b~mradermacher-i1",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "mistralai/Mistral-Medium-3.5-128B",
      "hub_repo": "mistralai/Mistral-Medium-3.5-128B",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 71427,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Medium-3.5-128B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-medium-3-5",
        "ids": [
          "mistralai/mistral-medium-3-5",
          "mistralai/mistral-medium-3-5:batch"
        ],
        "usd_per_m_in": 1.5,
        "usd_per_m_out": 7.5,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": 3.75,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 44.8263,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 134.1577,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 28.0218,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 13.4158,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 63.3042,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1173.6333,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 679.4719,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 32.4435,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Mistral-Small-4-119B-2603-GGUF:Q4_K_M",
      "family": "mistral-small-4",
      "member": "119b-2603",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "mistralai/Mistral-Small-4-119B-2603",
      "hub_repo": "mistralai/Mistral-Small-4-119B-2603",
      "publisher": "mradermacher",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 68815,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Small-4-119B-2603",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-small-2603",
        "ids": [
          "mistralai/mistral-small-2603"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": 0.015,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 42.6087,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 125.2441,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.81,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.8819,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.6158,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 60.2624,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1154.3092,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 668.2843,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 31.0389,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Mistral-Small-4-119B-2603-i1-GGUF:Q4_K_M",
      "family": "mistral-small-4",
      "member": "119b-2603~mradermacher-i1",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "mistralai/Mistral-Small-4-119B-2603",
      "hub_repo": "mistralai/Mistral-Small-4-119B-2603",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 68815,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Small-4-119B-2603",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-small-2603",
        "ids": [
          "mistralai/mistral-small-2603"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": 0.015,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 42.6087,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 125.2441,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.81,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.8819,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.6158,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 60.2624,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1154.3092,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 668.2843,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 31.0389,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/North-Mini-Code-1.0-i1-GGUF:IQ3_XXS",
      "family": "north-1",
      "member": "mini-code-1-0~mradermacher-i1",
      "quant": "IQ3_XXS",
      "engine": "ollama",
      "modality": "text",
      "base": "CohereLabs/North-Mini-Code-1.0",
      "hub_repo": "CohereLabs/North-Mini-Code-1.0",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 2,
      "disk_mib": 11338,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/CohereLabs/North-Mini-Code-1.0",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 18.4389
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 18.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.4284,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 2.15,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 21.835,
          "first_at": "2026-11-08T19:13:30.493Z",
          "last_at": "2026-11-08T19:13:30.493Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call"
          ],
          "usd_per_m_out": 18.4389,
          "usd_per_m_energy": 0.6065,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 120.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.0402,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 41.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.953,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.1622,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 116.2306,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 67.9653,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 61.05,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.3485,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/North-Mini-Code-1.0-i1-GGUF:Q4_K_M",
      "family": "north-1",
      "member": "mini-code-1-0~mradermacher-i1",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "CohereLabs/North-Mini-Code-1.0",
      "hub_repo": "CohereLabs/North-Mini-Code-1.0",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 17732,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/CohereLabs/North-Mini-Code-1.0",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.375,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 16.7966,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 76.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.386,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 26.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.686,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.6844,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 194.0967,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 112.3718,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 39.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.9334,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:IQ1_M",
      "family": "gpt-oss",
      "member": "20b~davidau-openai-claude-4-5-opus-heretic-uncensored",
      "quant": "IQ1_M",
      "engine": "ollama",
      "modality": "text",
      "base": "DavidAU/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
      "hub_repo": "DavidAU/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 5,
      "disk_mib": 11460,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/DavidAU/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 9.3955
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.4809,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": 4.23,
          "effective_ctx_best": null,
          "joules_per_token_best": 134.507,
          "first_at": "2026-11-09T17:51:27.318Z",
          "last_at": "2026-11-09T17:51:27.318Z",
          "caps": [],
          "usd_per_m_out": 9.3955,
          "usd_per_m_energy": 3.7363,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 118.95,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.0845,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 41.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9656,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.2454,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 118.0859,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 68.6311,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 60.4,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.397,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q3_K_S",
      "family": "gpt-oss",
      "member": "20b~davidau-openai-claude-4-5-opus-heretic-uncensored",
      "quant": "Q3_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "DavidAU/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
      "hub_repo": "DavidAU/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 2,
      "disk_mib": 11502,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/DavidAU/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 8.4273
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.4979,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 4.72,
          "effective_ctx_best": null,
          "joules_per_token_best": 13.746,
          "first_at": "2026-10-28T12:40:22.628Z",
          "last_at": "2026-10-28T12:40:22.628Z",
          "caps": [],
          "usd_per_m_out": 8.4273,
          "usd_per_m_energy": 0.3818,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 118.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.1,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 41.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9699,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.2737,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 118.1361,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 69.2522,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 60.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.4136,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q4_0",
      "family": "gpt-oss",
      "member": "20b~davidau-openai-claude-4-5-opus-heretic-uncensored",
      "quant": "Q4_0",
      "engine": "ollama",
      "modality": "text",
      "base": "DavidAU/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
      "hub_repo": "DavidAU/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 11585,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/DavidAU/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 8.2665
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.5347,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 4.82,
          "effective_ctx_best": null,
          "joules_per_token_best": 17.507,
          "first_at": "2026-10-28T13:21:07.008Z",
          "last_at": "2026-10-28T13:21:07.008Z",
          "caps": [],
          "usd_per_m_out": 8.2665,
          "usd_per_m_energy": 0.4863,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 117.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.1298,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 40.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9785,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.3306,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 120.0002,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 69.3104,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 59.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.4465,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/Qwen3-VL-8B-Instruct-abliterated-GGUF:Q2_K",
      "family": "qwen-3",
      "member": "vl-8b~prithivmlmods-abliterated",
      "quant": "Q2_K",
      "engine": "ollama",
      "modality": "text",
      "base": "prithivMLmods/Qwen3-VL-8B-Instruct-abliterated-v1",
      "hub_repo": "prithivMLmods/Qwen3-VL-8B-Instruct-abliterated-v1",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 5,
      "disk_mib": 3129,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/prithivMLmods/Qwen3-VL-8B-Instruct-abliterated-v1",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W35",
        "2026-W37"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "windows/amd64/intel/16g",
        "usd_per_m_out": 19.1455
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 65.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.12,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.1772,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 435.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0997,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 150.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2187,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 61.09,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8673,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 22.43,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-24T01:18:28.315Z",
          "last_at": "2026-08-27T07:24:23.270Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.ta",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.read",
            "vision.spatial"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 29.5255,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 3.84,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-12T03:12:02.558Z",
          "last_at": "2026-09-13T23:45:05.006Z",
          "caps": [
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.read"
          ],
          "usd_per_m_out": 19.1455,
          "usd_per_m_energy": 0.2532,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 3.59,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-10T17:40:01.299Z",
          "last_at": "2026-09-10T17:40:01.299Z",
          "caps": [
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "est_usd_per_m_out": 12.0061,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 221.22,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1699,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:F16",
      "family": "functiongemma",
      "member": "270m~muxodious-abominable-heresy",
      "quant": "F16",
      "engine": "ollama",
      "modality": "text",
      "base": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "hub_repo": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 4,
      "disk_mib": 517,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MuXodious/functiongemma-270m-it-abominable-heresy",
        "fetched_at": "2026-11-14T10:31:42.659Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 1.3579
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 397.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1797,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 22.71,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.38,
          "first_at": "2026-10-25T05:20:27.176Z",
          "last_at": "2026-10-25T05:20:27.176Z",
          "caps": [
            "lang.hi",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.3579,
          "usd_per_m_energy": 0.0661,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2636.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1809,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 912.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0338,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 369.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3028,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.7374,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 25.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7554,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1338.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1917,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:IQ4_XS",
      "family": "functiongemma",
      "member": "270m~muxodious-abominable-heresy",
      "quant": "IQ4_XS",
      "engine": "ollama",
      "modality": "text",
      "base": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "hub_repo": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 9,
      "disk_mib": 230,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MuXodious/functiongemma-270m-it-abominable-heresy",
        "fetched_at": "2026-11-14T10:31:42.659Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.7933
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 892.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0797,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 38.58,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.087,
          "first_at": "2026-10-25T02:43:33.986Z",
          "last_at": "2026-10-25T02:43:33.986Z",
          "caps": [
            "lang.bn",
            "lang.hi",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.7933,
          "usd_per_m_energy": 0.058,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5926.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0804,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2050.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0149,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 831.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1344,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 33.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.101,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 58.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.222,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3009.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0852,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q2_K",
      "family": "functiongemma",
      "member": "270m~muxodious-abominable-heresy",
      "quant": "Q2_K",
      "engine": "ollama",
      "modality": "text",
      "base": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "hub_repo": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 5,
      "disk_mib": 226,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MuXodious/functiongemma-270m-it-abominable-heresy",
        "fetched_at": "2026-11-14T10:31:42.659Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.8716
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 908.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0783,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 35.11,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.807,
          "first_at": "2026-10-25T02:32:25.632Z",
          "last_at": "2026-10-25T02:32:25.632Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.8716,
          "usd_per_m_energy": 0.0502,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6031.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.079,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2087.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0147,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 845.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1321,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 34.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.0643,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 59.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2007,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3062.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0837,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q3_K_S",
      "family": "functiongemma",
      "member": "270m~muxodious-abominable-heresy",
      "quant": "Q3_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "hub_repo": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 2,
      "disk_mib": 225,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MuXodious/functiongemma-270m-it-abominable-heresy",
        "fetched_at": "2026-11-14T10:31:42.659Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.8333
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 912.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.078,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 36.72,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.151,
          "first_at": "2026-10-25T02:23:43.185Z",
          "last_at": "2026-10-25T02:23:43.185Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.8333,
          "usd_per_m_energy": 0.0597,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6058.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0787,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2096.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0146,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 849.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1315,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 34.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.0553,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 59.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1953,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3076.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0834,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q4_K_M",
      "family": "functiongemma",
      "member": "270m~muxodious-abominable-heresy",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "hub_repo": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 241,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MuXodious/functiongemma-270m-it-abominable-heresy",
        "fetched_at": "2026-11-14T10:31:42.659Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9022
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 852.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0835,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 33.93,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.228,
          "first_at": "2026-10-25T03:13:57.494Z",
          "last_at": "2026-10-25T03:13:57.494Z",
          "caps": [
            "code.write",
            "lang.hi",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.9022,
          "usd_per_m_energy": 0.0619,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5656.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0843,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1957.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0156,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 793.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1409,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 32.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.2013,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 55.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2805,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2872.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0893,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q4_K_S",
      "family": "functiongemma",
      "member": "270m~muxodious-abominable-heresy",
      "quant": "Q4_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "hub_repo": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 238,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MuXodious/functiongemma-270m-it-abominable-heresy",
        "fetched_at": "2026-11-14T10:31:42.659Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9072
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 862.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0825,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 33.74,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.546,
          "first_at": "2026-10-25T02:55:20.924Z",
          "last_at": "2026-10-25T02:55:20.924Z",
          "caps": [
            "code.write",
            "lang.hi",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.9072,
          "usd_per_m_energy": 0.0429,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5727.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0832,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1981.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0154,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 803.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1391,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 32.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.1743,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 56.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2645,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2908.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0882,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/functiongemma-270m-it-abominable-heresy-GGUF:Q8_0",
      "family": "functiongemma",
      "member": "270m~muxodious-abominable-heresy",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "hub_repo": "MuXodious/functiongemma-270m-it-abominable-heresy",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 1,
      "disk_mib": 278,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MuXodious/functiongemma-270m-it-abominable-heresy",
        "fetched_at": "2026-11-14T10:31:42.659Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9519
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 738.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0964,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 32.19,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.7,
          "first_at": "2026-10-25T03:47:06.102Z",
          "last_at": "2026-10-25T03:47:06.102Z",
          "caps": [
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.9519,
          "usd_per_m_energy": 0.0472,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4903.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0972,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1696.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0181,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 687.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1625,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 28.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.541,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 48.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4778,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2489.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.103,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/mradermacher/granite-4.2-8b-i1-GGUF:Q4_K_M",
      "family": "granite-4-2",
      "member": "8b~mradermacher-i1",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "ibm-granite/granite-4.2-8b",
      "hub_repo": "ibm-granite/granite-4.2-8b",
      "publisher": "mradermacher",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 5100,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ibm-granite/granite-4.2-8b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "ibm-granite/granite-4.2-8b",
        "ids": [
          "ibm-granite/granite-4.2-8b"
        ],
        "usd_per_m_in": 0.06,
        "usd_per_m_out": 0.15,
        "usd_per_m_cached_in": 0.015,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 40.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8655,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.7209,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 267.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7983,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 92.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3737,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 37.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.0864,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 49.0997,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.63,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 28.5637,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 135.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.9187,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/ngquocvinh/EmbeddingGemma-2-GGUF:Q4_K_M",
      "family": "embeddinggemma-2",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/embeddinggemma-2",
      "hub_repo": "google/embeddinggemma-2",
      "publisher": "ngquocvinh",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 173,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/embeddinggemma-2",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1187.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0599,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 273.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1116,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7879.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0605,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2726.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0112,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-06T12:40:34.040Z",
          "last_at": "2026-11-06T12:40:34.040Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 1104.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1011,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 45.03,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.5792,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 77.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9185,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4001.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0641,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/notSnix/Step-3.7-Flash-Q4_K_M-MTP-GGUF:Q4_K_M",
      "family": "step-3-7",
      "member": "flash~notsnix-k-m-mtp",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "stepfun-ai/Step-3.7-Flash",
      "hub_repo": "stepfun-ai/Step-3.7-Flash",
      "publisher": "notSnix",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 113089,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "stepfun/step-3.7-flash",
        "ids": [
          "stepfun/step-3.7-flash"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 1.15,
        "usd_per_m_cached_in": 0.04,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 89.5628,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 291.5446,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.05,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 47.1834,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 29.3642,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 120.4496,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2328.6327,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1358.3691,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 56.9088,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/openbmb/MiniCPM-V-4-gguf:Q4_K_M",
      "family": "minicpm-v-4",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "openbmb/MiniCPM-V-4",
      "hub_repo": "openbmb/MiniCPM-V-4",
      "publisher": "openbmb",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 2088,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openbmb/MiniCPM-V-4",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 1.1708
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 98.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7388,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4154,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 652.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7325,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 225.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1422,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 96.71,
          "effective_ctx_best": null,
          "joules_per_token_best": 4.724,
          "first_at": "2026-11-06T10:45:43.688Z",
          "last_at": "2026-11-06T10:45:43.688Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.1708,
          "usd_per_m_energy": 0.1312,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 3.73,
          "effective_ctx_best": null,
          "joules_per_token_best": 50.705,
          "first_at": "2026-10-28T14:23:26.586Z",
          "last_at": "2026-10-28T14:23:26.586Z",
          "caps": [
            "output.schema",
            "structured_output",
            "vision",
            "vision.count",
            "vision.read"
          ],
          "usd_per_m_out": 19.483,
          "usd_per_m_energy": 1.4085,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 3.21,
          "effective_ctx_best": null,
          "joules_per_token_best": 72.671,
          "first_at": "2026-11-07T22:45:53.637Z",
          "last_at": "2026-11-07T22:45:53.637Z",
          "caps": [
            "output.schema",
            "structured_output",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 22.6391,
          "usd_per_m_energy": 2.0186,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 331.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7781,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/openbmb/MiniCPM-V-4.6-gguf:Q4_K_M",
      "family": "minicpm-v-4-6",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "openbmb/MiniCPM-V-4.6",
      "hub_repo": "openbmb/MiniCPM-V-4.6",
      "publisher": "openbmb",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 504,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openbmb/MiniCPM-V-4.6",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 0.8551
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 407.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1752,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 93.99,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.328,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2704.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1763,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 935.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0329,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 130.91,
          "effective_ctx_best": 262144,
          "joules_per_token_best": 2.091,
          "first_at": "2026-11-07T03:31:45.467Z",
          "last_at": "2026-11-07T03:31:45.467Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.8551,
          "usd_per_m_energy": 0.0581,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.6172,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": 27.29,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 2.985,
          "first_at": "2026-11-10T23:56:39.786Z",
          "last_at": "2026-11-10T23:56:39.786Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call"
          ],
          "usd_per_m_out": 2.6157,
          "usd_per_m_energy": 0.0829,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1373.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1869,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q4_K_M",
      "family": "minicpm-v-4-7",
      "member": "1b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "openbmb/MiniCPM-V-4.7-1B",
      "hub_repo": "openbmb/MiniCPM-V-4.7-1B",
      "publisher": "openbmb",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 640,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openbmb/MiniCPM-V-4.7-1B",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 0.5067
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 320.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2228,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 74.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.418,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2129.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.224,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 737.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.042,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 221.15,
          "effective_ctx_best": 262144,
          "joules_per_token_best": 9.014,
          "first_at": "2026-11-06T12:49:22.462Z",
          "last_at": "2026-11-06T12:49:22.462Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.5067,
          "usd_per_m_energy": 0.2504,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.8745,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 20.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.4158,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1081.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2374,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/openbmb/MiniCPM-V-4.7-1B-gguf:Q8_0",
      "family": "minicpm-v-4-7",
      "member": "1b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "openbmb/MiniCPM-V-4.7-1B",
      "hub_repo": "openbmb/MiniCPM-V-4.7-1B",
      "publisher": "openbmb",
      "derivative": false,
      "cohort": 1,
      "rung": 1,
      "disk_mib": 1031,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openbmb/MiniCPM-V-4.7-1B",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 0.6046
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 199.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3605,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 14.97,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 2.226,
          "first_at": "2026-11-05T07:36:36.145Z",
          "last_at": "2026-11-05T07:36:36.145Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 2.0879,
          "usd_per_m_energy": 0.0618,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1322.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.361,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 457.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0683,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 185.86,
          "effective_ctx_best": 262144,
          "joules_per_token_best": 9.714,
          "first_at": "2026-11-06T08:06:44.658Z",
          "last_at": "2026-11-06T08:06:44.658Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.ta",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.6046,
          "usd_per_m_energy": 0.2698,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.4988,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.99,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.5282,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 671.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3829,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/openbmb/MiniCPM-V-4_5-gguf:Q4_K_M",
      "family": "minicpm-v-4-5",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "openbmb/MiniCPM-V-4_5",
      "hub_repo": "openbmb/MiniCPM-V-4_5",
      "publisher": "openbmb",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 4793,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openbmb/MiniCPM-V-4_5",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 10.1968
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 42.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7473,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.4734,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 284.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6892,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 98.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3487,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 11.32,
          "effective_ctx_best": 16384,
          "joules_per_token_best": 9.522,
          "first_at": "2026-11-06T11:23:38.340Z",
          "last_at": "2026-11-06T11:23:38.340Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 10.1968,
          "usd_per_m_energy": 0.2645,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.63,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 45.9342,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.8361,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 144.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8013,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/openbmb/MiniCPM-o-4_5-gguf:Q4_K_M",
      "family": "minicpm-o-4-5",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "openbmb/MiniCPM-o-4_5",
      "hub_repo": "openbmb/MiniCPM-o-4_5",
      "publisher": "openbmb",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 4793,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openbmb/MiniCPM-o-4_5",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 5.7058
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 42.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7473,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.4734,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 284.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6892,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 98.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3487,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 20.23,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 7.387,
          "first_at": "2026-11-06T09:57:23.685Z",
          "last_at": "2026-11-06T09:57:23.685Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call"
          ],
          "usd_per_m_out": 5.7058,
          "usd_per_m_energy": 0.2052,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.63,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 45.9342,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.8361,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 144.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8013,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/openbmb/MiniCPM4.1-8B-GGUF:Q4_K_M",
      "family": "minicpm-4-1",
      "member": "8b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "openbmb/MiniCPM4.1-8B-GGUF",
      "hub_repo": "openbmb/MiniCPM4.1-8B-GGUF",
      "publisher": "openbmb",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 4735,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openbmb/MiniCPM4.1-8B-GGUF",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 2.5187
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.37,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7253,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.427,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 287.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6686,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 99.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.344,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 45.81,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 3.049,
          "first_at": "2026-11-06T22:53:16.931Z",
          "last_at": "2026-11-06T22:53:16.931Z",
          "caps": [
            "code.reason",
            "lang.en",
            "lang.hi",
            "lang.ml",
            "lang.ta",
            "lang.te",
            "output.long",
            "reason.channel",
            "research.synthesize",
            "thinking"
          ],
          "usd_per_m_out": 2.5187,
          "usd_per_m_energy": 0.0847,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 45.3488,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.4401,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 146.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7792,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/openbmb/MiniCPM5-1B-GGUF:Q4_K_M",
      "family": "minicpm-5",
      "member": "1b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "openbmb/MiniCPM5-1B-GGUF",
      "hub_repo": "openbmb/MiniCPM5-1B-GGUF",
      "publisher": "openbmb",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 656,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openbmb/MiniCPM5-1B-GGUF",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 0.538
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 313.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2284,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 72.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4286,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2077.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2296,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 719.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.043,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 208.3,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 1.39,
          "first_at": "2026-11-06T09:46:25.417Z",
          "last_at": "2026-11-06T09:46:25.417Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 0.538,
          "usd_per_m_energy": 0.0386,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.019,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W46"
          ],
          "tokens_per_sec_best": 21.73,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 2.338,
          "first_at": "2026-10-26T19:05:11.691Z",
          "last_at": "2026-11-11T06:00:39.094Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 3.2907,
          "usd_per_m_energy": 0.0649,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 381.36,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 1.79,
          "first_at": "2026-11-08T13:20:06.835Z",
          "last_at": "2026-11-08T13:20:06.835Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 0.6733,
          "usd_per_m_energy": 0.0497,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/patrickbdevaney/GLM-5.3-Flash-REAP50-GGUF:Q4_K_M",
      "family": "glm-5-3",
      "member": "flash~patrickbdevaney-reap50",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "zai-org/GLM-5.3-Flash",
      "hub_repo": "zai-org/GLM-5.3-Flash",
      "publisher": "patrickbdevaney",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 94729,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.3-Flash",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.3-flash",
        "ids": [
          "z-ai/glm-5.3-flash",
          "z-ai/glm-5.3-flash:batch"
        ],
        "usd_per_m_in": 0.0399,
        "usd_per_m_out": 0.0784,
        "usd_per_m_cached_in": 0.03,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.18,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 68.2317,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 215.0146,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 14.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 38.4724,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 21.5878,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 93.3755,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1850.786,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1057.592,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 45.6006,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/peculiar-ragdoll/Dirk-Qwen3.8-27B-GGUF:Q4_K_M",
      "family": "qwen-3-8",
      "member": "27b~peculiar-ragdoll-dirk",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "peculiar-ragdoll",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 15701,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 4.3368,
        "vs_api": 2.31
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.4029,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.3027,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 86.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.6359,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 9.96,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 15.717,
          "first_at": "2026-10-28T09:31:59.554Z",
          "last_at": "2026-10-28T09:31:59.554Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 4.3368,
          "usd_per_m_energy": 0.4366,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.2141,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 167.4995,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 98.5291,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 44.09,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.1018,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/sabafallah/DeepSeek-OCR-2-GGUF:Q4_K_M",
      "family": "deepseek-ocr-2",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-OCR-2",
      "hub_repo": "deepseek-ai/DeepSeek-OCR-2",
      "publisher": "sabafallah",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 1859,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-OCR-2",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 0.3376
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 110.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6561,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 25.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2531,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 733.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6519,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 253.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1258,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 334.82,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.393,
          "first_at": "2026-11-06T10:40:24.795Z",
          "last_at": "2026-11-06T10:40:24.795Z",
          "caps": [
            "code.reason",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.3376,
          "usd_per_m_energy": 0.0665,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 17.2995,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 15.36,
          "effective_ctx_best": null,
          "joules_per_token_best": 15.414,
          "first_at": "2026-11-07T22:09:04.247Z",
          "last_at": "2026-11-07T22:09:04.247Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 4.7191,
          "usd_per_m_energy": 0.4282,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": 411.73,
          "effective_ctx_best": null,
          "joules_per_token_best": 3.969,
          "first_at": "2026-11-10T09:56:54.990Z",
          "last_at": "2026-11-10T09:56:54.990Z",
          "caps": [
            "code.reason",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.626,
          "usd_per_m_energy": 0.1103,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/sabafallah/DeepSeek-OCR-GGUF:Q4_K_M",
      "family": "deepseek-ocr",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-OCR",
      "hub_repo": "deepseek-ai/DeepSeek-OCR",
      "publisher": "sabafallah",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 1859,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-OCR",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 0.3323
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 110.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6561,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 25.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2531,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 733.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6519,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 253.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1258,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 340.22,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.492,
          "first_at": "2026-11-06T10:35:49.206Z",
          "last_at": "2026-11-06T10:35:49.206Z",
          "caps": [
            "lang.en",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 0.3323,
          "usd_per_m_energy": 0.0692,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 17.2995,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 13.32,
          "effective_ctx_best": null,
          "joules_per_token_best": 23.379,
          "first_at": "2026-11-07T21:26:21.283Z",
          "last_at": "2026-11-07T21:26:21.283Z",
          "caps": [
            "lang.en"
          ],
          "usd_per_m_out": 5.4418,
          "usd_per_m_energy": 0.6494,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 372.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6923,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/second-state/embeddinggemma-300m-GGUF:Q2_K",
      "family": "embeddinggemma",
      "member": "300m",
      "quant": "Q2_K",
      "engine": "ollama",
      "modality": "text",
      "base": "google/embeddinggemma-300m",
      "hub_repo": "google/embeddinggemma-300m",
      "publisher": "second-state",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 202,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/embeddinggemma-300m",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1016.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.07,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 234.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1304,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6748.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0706,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2335.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0131,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 946.25,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.118,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 38.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8444,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-14T01:33:03.701Z",
          "last_at": "2026-11-14T04:18:38.644Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 66.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0728,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3426.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0748,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/second-state/embeddinggemma-300m-GGUF:Q3_K_S",
      "family": "embeddinggemma",
      "member": "300m",
      "quant": "Q3_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "google/embeddinggemma-300m",
      "hub_repo": "google/embeddinggemma-300m",
      "publisher": "second-state",
      "derivative": false,
      "cohort": 1,
      "rung": 2,
      "disk_mib": 208,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/embeddinggemma-300m",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 987.37,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0721,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 227.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1343,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6553.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0727,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2267.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0135,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 918.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1215,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 37.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8996,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-14T02:46:05.389Z",
          "last_at": "2026-11-14T04:18:35.022Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 64.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1049,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3327.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0771,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/second-state/embeddinggemma-300m-GGUF:Q4_K_M",
      "family": "embeddinggemma",
      "member": "300m",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/embeddinggemma-300m",
      "hub_repo": "google/embeddinggemma-300m",
      "publisher": "second-state",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 225,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/embeddinggemma-300m",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 912.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.078,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 210.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1453,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6058.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0787,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2096.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0146,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-06T12:39:41.676Z",
          "last_at": "2026-11-06T12:39:41.676Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 849.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1315,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 34.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.0553,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-02T21:10:05.554Z",
          "last_at": "2026-11-15T04:08:04.060Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 59.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1953,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-08T15:41:43.747Z",
          "last_at": "2026-11-10T09:57:02.153Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 3076.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0834,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/taronaeo/Qwen3.6-35B-A3B-BE-GGUF:Q4_K_M",
      "family": "qwen-3-6",
      "member": "35b-a3b~taronaeo-be",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.6-35B-A3B",
      "hub_repo": "Qwen/Qwen3.6-35B-A3B",
      "publisher": "taronaeo",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 20707,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.6-35B-A3B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.6-35b-a3b",
        "ids": [
          "qwen/qwen3.6-35b-a3b"
        ],
        "usd_per_m_in": 0.1,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.025,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 115.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.759,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 26.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.771,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 768.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6424,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 265.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1779,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 107.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1921,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 20.0054,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.6323,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 389.99,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7003,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/tencent/HY-MT1.5-1.8B-GGUF:Q4_K_M",
      "family": "hy-mt-1-5",
      "member": "1.8b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "tencent/HY-MT1.5-1.8B-GGUF",
      "hub_repo": "tencent/HY-MT1.5-1.8B-GGUF",
      "publisher": "tencent",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 1080,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/HY-MT1.5-1.8B-GGUF",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 2.4179
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 190.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3778,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7135,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1262.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3782,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 436.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0717,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 46.49,
          "effective_ctx_best": null,
          "joules_per_token_best": 236.861,
          "first_at": "2026-11-06T15:04:07.200Z",
          "last_at": "2026-11-06T15:04:07.200Z",
          "caps": [],
          "usd_per_m_out": 2.4179,
          "usd_per_m_energy": 6.5795,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.9655,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": 7.41,
          "effective_ctx_best": null,
          "joules_per_token_best": 222.706,
          "first_at": "2026-11-10T22:07:22.308Z",
          "last_at": "2026-11-10T22:07:22.308Z",
          "caps": [],
          "usd_per_m_out": 9.6965,
          "usd_per_m_energy": 6.1863,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 640.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4012,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_M",
      "family": "hy-mt-2",
      "member": "1.8b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "tencent/Hy-MT2-1.8B",
      "hub_repo": "tencent/Hy-MT2-1.8B",
      "publisher": "tencent",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 1080,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy-MT2-1.8B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": {
        "openrouter_id": "tencent/hy-mt2-1.8b",
        "ids": [
          "tencent/hy-mt2-1.8b"
        ],
        "usd_per_m_in": 0.044,
        "usd_per_m_out": 0.177,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "windows/amd64/nvidia/16g",
        "usd_per_m_out": 1.102,
        "vs_api": 6.23
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 190.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3778,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7135,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1262.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3782,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 436.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0717,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 36.5,
          "effective_ctx_best": null,
          "joules_per_token_best": 3.334,
          "first_at": "2026-11-06T10:30:48.157Z",
          "last_at": "2026-11-06T10:30:48.157Z",
          "caps": [],
          "usd_per_m_out": 3.0796,
          "usd_per_m_energy": 0.0926,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.9655,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": 14.17,
          "effective_ctx_best": null,
          "joules_per_token_best": 21.081,
          "first_at": "2026-11-07T14:39:51.356Z",
          "last_at": "2026-11-14T04:08:54.697Z",
          "caps": [],
          "usd_per_m_out": 5.0707,
          "usd_per_m_energy": 0.5856,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W45"
          ],
          "tokens_per_sec_best": 233.33,
          "effective_ctx_best": null,
          "joules_per_token_best": 397.589,
          "first_at": "2026-10-29T15:58:38.778Z",
          "last_at": "2026-11-08T13:20:44.172Z",
          "caps": [],
          "usd_per_m_out": 1.102,
          "usd_per_m_energy": 11.0441,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        }
      }
    },
    {
      "ref": "hf.co/tencent/Hy-MT2-30B-A3B-GGUF:Q4_K_M",
      "family": "hy-mt-2",
      "member": "30b-a3b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "tencent/Hy-MT2-30B-A3B-GGUF",
      "hub_repo": "tencent/Hy-MT2-30B-A3B-GGUF",
      "publisher": "tencent",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 17391,
      "active_share": 0.1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy-MT2-30B-A3B-GGUF",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "tencent/hy-mt2-30b-a3b",
        "ids": [
          "tencent/hy-mt2-30b-a3b"
        ],
        "usd_per_m_in": 0.074,
        "usd_per_m_out": 0.295,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 8.3152,
        "vs_api": 28.19
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 118.09,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7208,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 5.36,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 53.363,
          "first_at": "2026-10-27T17:52:47.768Z",
          "last_at": "2026-10-27T19:09:54.619Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 8.3152,
          "usd_per_m_energy": 1.4823,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 783.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.626,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 271.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1643,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 109.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1435,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19.0011,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.0552,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 398.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6794,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/tencent/Hy-MT2-7B-GGUF:Q4_K_M",
      "family": "hy-mt-2",
      "member": "7b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "tencent/Hy-MT2-7B",
      "hub_repo": "tencent/Hy-MT2-7B",
      "publisher": "tencent",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 4410,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy-MT2-7B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy-mt2-7b",
        "ids": [
          "tencent/hy-mt2-7b"
        ],
        "usd_per_m_in": 0.074,
        "usd_per_m_out": 0.295,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 46.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6011,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.1663,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 309.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.5532,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 106.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3179,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.6561,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 42.1249,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 24.5267,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 156.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6554,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/turintech/Qwen3.5-9B-Artemis-GGUF:Q4_K_M",
      "family": "qwen-3-5",
      "member": "9b~turintech-artemis",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.5-9B",
      "hub_repo": "Qwen/Qwen3.5-9B",
      "publisher": "turintech",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 5398,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.5-9B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.5-9b",
        "ids": [
          "qwen/qwen3.5-9b"
        ],
        "usd_per_m_in": 0.037,
        "usd_per_m_out": 0.054,
        "usd_per_m_cached_in": 0.033,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 38.05,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.9807,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.9646,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 252.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.9044,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 87.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3984,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 35.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.2737,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 52.3369,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 30.3892,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 128.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.0326,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/GLM-4.7-Flash-GGUF:Q4_K_M",
      "family": "glm-4-7",
      "member": "flash",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "zai-org/GLM-4.7-Flash",
      "hub_repo": "zai-org/GLM-4.7-Flash",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 17464,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.7-Flash",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "z-ai/glm-4.7-flash",
        "ids": [
          "z-ai/glm-4.7-flash"
        ],
        "usd_per_m_in": 0.0605,
        "usd_per_m_out": 0.4,
        "usd_per_m_cached_in": null,
        "context_length": 200000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.9922,
        "vs_api": 2.48
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.2436,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W44",
            "2026-W45"
          ],
          "tokens_per_sec_best": 3.55,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 9.384,
          "first_at": "2026-10-27T16:11:19.116Z",
          "last_at": "2026-11-04T10:14:53.192Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.es",
            "lang.hi",
            "lang.ml",
            "lang.mr",
            "lang.ne",
            "lang.nl",
            "lang.or",
            "lang.pa",
            "lang.ru",
            "lang.ta",
            "lang.te",
            "lang.ur",
            "output.long",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 12.5715,
          "usd_per_m_energy": 0.2607,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 78.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.2867,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 44.98,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 3.68,
          "first_at": "2026-11-03T03:36:25.966Z",
          "last_at": "2026-11-03T03:36:25.966Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 0.9922,
          "usd_per_m_energy": 0.1022,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 40.58,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 2.601,
          "first_at": "2026-11-07T04:19:50.619Z",
          "last_at": "2026-11-07T04:19:50.619Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 3.0986,
          "usd_per_m_energy": 0.0722,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 189.2988,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 110.6292,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 39.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.823,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/LFM2.5-1.2B-Instruct-GGUF:Q4_K_M",
      "family": "lfm-2-5",
      "member": "1.2b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "LiquidAI/LFM2.5-1.2B-Instruct",
      "hub_repo": "LiquidAI/LFM2.5-1.2B-Instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 697,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "lfm1.0",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/LiquidAI/LFM2.5-1.2B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 1.9487
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 294.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2428,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 15.9,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 2.84,
          "first_at": "2026-10-24T23:37:19.190Z",
          "last_at": "2026-10-24T23:37:19.190Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 1.9487,
          "usd_per_m_energy": 0.0789,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1955.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2439,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 676.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0458,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 274.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4087,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.3989,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.22,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.7221,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 993.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2586,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/LFM2.5-24B-A2B-GGUF:IQ3_XXS",
      "family": "lfm-2-5",
      "member": "24b-a2b",
      "quant": "IQ3_XXS",
      "engine": "ollama",
      "modality": "text",
      "base": "LiquidAI/LFM2.5-24B-A2B",
      "hub_repo": "LiquidAI/LFM2.5-24B-A2B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 2,
      "disk_mib": 8544,
      "active_share": 0.0833,
      "licence": {
        "id": "other",
        "name": "lfm1.0",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/LiquidAI/LFM2.5-24B-A2B",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": {
        "openrouter_id": "liquid/lfm-2.5-24b-a2b",
        "ids": [
          "liquid/lfm-2.5-24b-a2b"
        ],
        "usd_per_m_in": 0.05,
        "usd_per_m_out": 0.7,
        "usd_per_m_cached_in": 0.025,
        "context_length": 128000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 10.2664,
        "vs_api": 14.67
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 288.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2702,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 3.64,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 19.181,
          "first_at": "2026-10-30T17:01:31.059Z",
          "last_at": "2026-10-30T17:01:31.059Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 10.2664,
          "usd_per_m_energy": 0.5328,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1914.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2525,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 662.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0564,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 268.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4413,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.123,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 18.81,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.1428,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 972.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2707,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:F16",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "F16",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-3B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-3B-Instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 4,
      "disk_mib": 6135,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W38"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "windows/amd64/nvidia/16g",
        "usd_per_m_out": 2.7531,
        "vs_api": 8.22
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 33.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.269,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.5867,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 222.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.1672,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 76.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4605,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 31.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.7394,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 59.8149,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 34.8463,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 94.89,
          "effective_ctx_best": 16384,
          "joules_per_token_best": null,
          "first_at": "2026-09-16T18:26:43.629Z",
          "last_at": "2026-09-16T18:26:43.629Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.mr",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 2.7531,
          "usd_per_m_energy": 0.0659,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ1_S",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "IQ1_S",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-3B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-3B-Instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 870,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W38",
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.3295,
        "vs_api": 0.98
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 236.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3037,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 54.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5716,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1566.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3046,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 94.46,
          "effective_ctx_best": null,
          "joules_per_token_best": 46.86,
          "first_at": "2026-10-20T11:20:10.193Z",
          "last_at": "2026-10-20T11:20:10.193Z",
          "caps": [
            "lang.en"
          ],
          "usd_per_m_out": 0.3295,
          "usd_per_m_energy": 1.3017,
          "tariff_basis": "bootstrap",
          "verdict": "cheaper than the API"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 219.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5109,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.95,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.009,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.4,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.6546,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 225.19,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T16:01:57.006Z",
          "last_at": "2026-09-15T16:01:57.006Z",
          "caps": [
            "lang.en"
          ],
          "usd_per_m_out": 1.1411,
          "usd_per_m_energy": 0.0278,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ2_XXS",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "IQ2_XXS",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-3B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-3B-Instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 998,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W38"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "windows/amd64/nvidia/16g",
        "usd_per_m_out": 2.0702,
        "vs_api": 6.18
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 205.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3488,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 47.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.658,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1365.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3495,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 472.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0661,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 191.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5865,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.81,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.1913,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.3491,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 124.17,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T16:05:28.433Z",
          "last_at": "2026-09-15T16:05:28.433Z",
          "caps": [
            "code.reason",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 2.0702,
          "usd_per_m_energy": 0.0503,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:IQ3_XXS",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "IQ3_XXS",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-3B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-3B-Instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 2,
      "disk_mib": 1306,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W38"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "windows/amd64/nvidia/16g",
        "usd_per_m_out": 1.2897,
        "vs_api": 3.85
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 157.25,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4581,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 36.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8679,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1043.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4576,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 361.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0872,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 146.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7693,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.97,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.0662,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.021,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 199.51,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T16:27:15.526Z",
          "last_at": "2026-09-15T16:27:15.526Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.mr",
            "lang.pa",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 1.2897,
          "usd_per_m_energy": 0.0313,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_0",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "Q4_0",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-3B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-3B-Instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 3,
      "disk_mib": 1832,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W38"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "windows/amd64/nvidia/16g",
        "usd_per_m_out": 1.2569,
        "vs_api": 3.75
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 112.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6464,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 25.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2339,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 744.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6424,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 257.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1239,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 104.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0832,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.25,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 17.0501,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.9129,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 205.06,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T16:49:08.633Z",
          "last_at": "2026-09-15T16:49:08.633Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 1.2569,
          "usd_per_m_energy": 0.0305,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q4_K_M",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-3B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-3B-Instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 1925,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W38",
        "2026-W45",
        "2026-W46"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 1.161,
        "vs_api": 3.47
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 106.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6799,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 24.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2996,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 708.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6752,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 245.03,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1305,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 97.41,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.199,
          "first_at": "2026-11-06T23:21:39.586Z",
          "last_at": "2026-11-06T23:21:39.586Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.ml",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 1.161,
          "usd_per_m_energy": 0.0333,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.05,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 17.9108,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": 7.66,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 11.417,
          "first_at": "2026-11-07T14:50:57.406Z",
          "last_at": "2026-11-14T04:24:09.329Z",
          "caps": [
            "faithful_tool_result",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 9.4698,
          "usd_per_m_energy": 0.3171,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 203.94,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T14:50:29.236Z",
          "last_at": "2026-09-15T14:50:29.236Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 1.2642,
          "usd_per_m_energy": 0.0306,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Llama-3.2-3B-Instruct-GGUF:Q8_0",
      "family": "llama-3-2",
      "member": "3b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "meta-llama/Llama-3.2-3B-Instruct",
      "hub_repo": "meta-llama/Llama-3.2-3B-Instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 1,
      "disk_mib": 3263,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-llama/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W38"
      ],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "windows/amd64/nvidia/16g",
        "usd_per_m_out": 3.5559,
        "vs_api": 10.61
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 62.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1698,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 14.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.2777,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 417.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.147,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 144.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2288,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 58.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.9492,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 30.8066,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 17.958,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 72.81,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T17:42:09.844Z",
          "last_at": "2026-09-15T17:42:09.844Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 3.5559,
          "usd_per_m_energy": 0.0858,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Magistral-Small-2506-GGUF:Q4_K_M",
      "family": "magistral-small",
      "member": "2506",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "mistralai/Magistral-Small-2506",
      "hub_repo": "mistralai/Magistral-Small-2506",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 13669,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Magistral-Small-2506",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 22.3817
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.4658,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.9713,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 99.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.8897,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 34.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2037,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 5.48,
          "effective_ctx_best": 40960,
          "joules_per_token_best": 9.679,
          "first_at": "2026-11-06T18:57:28.546Z",
          "last_at": "2026-11-06T18:57:28.546Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 22.3817,
          "usd_per_m_energy": 0.2689,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 144.0282,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 83.7715,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 50.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.2799,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/North-Mini-Code-1.0-GGUF:Q4_K_M",
      "family": "north-1",
      "member": "mini-code-1-0",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "CohereLabs/North-Mini-Code-1.0",
      "hub_repo": "CohereLabs/North-Mini-Code-1.0",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 18313,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/CohereLabs/North-Mini-Code-1.0",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 14.1624
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.6606,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 3.2,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 27.161,
          "first_at": "2026-10-27T18:53:13.545Z",
          "last_at": "2026-10-27T20:26:54.884Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call"
          ],
          "usd_per_m_out": 14.1624,
          "usd_per_m_energy": 0.7545,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 74.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.6017,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 25.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7593,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.1102,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 199.7102,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 117.6375,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 37.8,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.1734,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:BF16",
      "family": "phi-4",
      "member": "mini",
      "quant": "BF16",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 4,
      "disk_mib": 7324,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W38",
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 1.1429
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 28.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7437,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.6224,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 186.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.5923,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 31.83,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 10.965,
          "first_at": "2026-10-22T09:33:40.320Z",
          "last_at": "2026-10-22T09:33:40.320Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.ml",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.1429,
          "usd_per_m_energy": 0.3046,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 26.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.5015,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 72.5778,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 42.0396,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 0.08,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T18:07:27.376Z",
          "last_at": "2026-09-15T18:07:27.376Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 3277.6286,
          "usd_per_m_energy": 78.125,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q2_K",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q2_K",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 1604,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.4624
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 128.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5645,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 29.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0742,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 849.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5623,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 68.61,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.538,
          "first_at": "2026-10-20T11:54:49.141Z",
          "last_at": "2026-10-20T11:54:49.141Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.4624,
          "usd_per_m_energy": 0.0705,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 119.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9468,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.8719,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.35,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.656,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 431.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5968,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q3_K_M",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q3_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 2,
      "disk_mib": 2019,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.4816
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 101.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7139,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 23.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.3666,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 675.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7082,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 66.57,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 11.782,
          "first_at": "2026-10-20T12:11:09.085Z",
          "last_at": "2026-10-20T12:11:09.085Z",
          "caps": [
            "code.write",
            "lang.bn",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.4816,
          "usd_per_m_energy": 0.3273,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 94.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1954,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 18.8123,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.63,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.9525,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 342.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7522,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q4_K_M",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 2376,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W35",
        "2026-W43",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.4316
      },
      "classes": {
        "darwin/amd64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 2.18,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-28T06:50:35.623Z",
          "last_at": "2026-08-28T06:50:35.623Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "est_usd_per_m_out": 19.4904,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 86.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8434,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6224,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 573.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.834,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 74.95,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 7.69,
          "first_at": "2026-10-21T21:01:20.128Z",
          "last_at": "2026-10-21T21:01:20.128Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 0.4316,
          "usd_per_m_energy": 0.2136,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 79.36,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.629,
          "first_at": "2026-11-06T19:21:02.380Z",
          "last_at": "2026-11-06T19:21:02.380Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.4297,
          "usd_per_m_energy": 0.0452,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 22.2274,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.9266,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 291.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8862,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q5_K_M",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q5_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 3,
      "disk_mib": 2716,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.4786
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 75.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9678,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8708,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 501.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9539,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 68.17,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 8.996,
          "first_at": "2026-10-22T08:56:06.701Z",
          "last_at": "2026-10-22T08:56:06.701Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.ml",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 0.4786,
          "usd_per_m_energy": 0.2499,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 70.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6161,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 25.4991,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.8443,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 254.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0141,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Phi-4-mini-instruct-GGUF:Q8_0",
      "family": "phi-4",
      "member": "mini",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 1,
      "disk_mib": 3895,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.6306
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 52.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4061,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.762,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 349.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.3706,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 53.26,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 12.377,
          "first_at": "2026-10-22T09:12:21.715Z",
          "last_at": "2026-10-22T09:12:21.715Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.ta",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 0.6306,
          "usd_per_m_energy": 0.3438,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 49.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.3374,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 37.071,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 21.5529,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 177.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4598,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ1_S",
      "family": "phi-4",
      "member": "mini-reasoning",
      "quant": "IQ1_S",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-reasoning",
      "hub_repo": "microsoft/Phi-4-mini-reasoning",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 5,
      "disk_mib": 1088,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-reasoning",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W34",
        "2026-W35",
        "2026-W37",
        "2026-W38",
        "2026-W41",
        "2026-W42"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "windows/amd64/nvidia/16g",
        "usd_per_m_out": 1.3614
      },
      "classes": {
        "darwin/amd64/apple/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-13T03:45:25.064Z",
          "last_at": "2026-09-13T03:45:25.064Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W42"
          ],
          "tokens_per_sec_best": 17.18,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-16T07:05:05.022Z",
          "last_at": "2026-10-16T07:05:05.022Z",
          "caps": [
            "lang.en"
          ],
          "usd_per_m_out": 4.1826,
          "usd_per_m_energy": 0.0566,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "darwin/arm64/apple/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W41"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-08T10:28:59.157Z",
          "last_at": "2026-10-08T10:28:59.157Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "darwin/arm64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 16.13,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-07T07:58:40.044Z",
          "last_at": "2026-09-07T07:58:40.044Z",
          "caps": [],
          "est_usd_per_m_out": 2.5692,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7189,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1252.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.381,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 433.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0722,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 175.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6399,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W34",
            "2026-W35"
          ],
          "tokens_per_sec_best": 27.82,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-23T16:21:43.845Z",
          "last_at": "2026-08-24T07:05:43.749Z",
          "caps": [
            "lang.en"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T07:42:59.096Z",
          "last_at": "2026-09-15T07:42:59.096Z",
          "caps": [],
          "est_tokens_per_sec": 7.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.036,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 5.6,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-13T02:05:19.379Z",
          "last_at": "2026-09-13T02:05:19.379Z",
          "caps": [],
          "usd_per_m_out": 12.8317,
          "usd_per_m_energy": 0.1736,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 5.6,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-11T04:54:16.851Z",
          "last_at": "2026-09-11T04:54:16.851Z",
          "caps": [],
          "est_usd_per_m_out": 7.4002,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37",
            "2026-W38",
            "2026-W41"
          ],
          "tokens_per_sec_best": 188.88,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-11T18:33:38.903Z",
          "last_at": "2026-10-07T17:03:57.382Z",
          "caps": [
            "lang.en"
          ],
          "usd_per_m_out": 1.3614,
          "usd_per_m_energy": 0.0331,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:IQ3_XXS",
      "family": "phi-4",
      "member": "mini-reasoning",
      "quant": "IQ3_XXS",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-reasoning",
      "hub_repo": "microsoft/Phi-4-mini-reasoning",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 2,
      "disk_mib": 1591,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-reasoning",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W37",
        "2026-W38",
        "2026-W41",
        "2026-W42"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "windows/amd64/nvidia/16g",
        "usd_per_m_out": 1.3596
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W42"
          ],
          "tokens_per_sec_best": 25.64,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-10-16T07:29:07.951Z",
          "last_at": "2026-10-16T07:29:07.951Z",
          "caps": [
            "code.reason",
            "lang.en",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking"
          ],
          "usd_per_m_out": 2.8185,
          "usd_per_m_energy": 0.0379,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "darwin/arm64/apple/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W41"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-08T11:03:39.716Z",
          "last_at": "2026-10-08T11:03:39.716Z",
          "caps": [
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 29.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0652,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 856.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5577,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 296.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.107,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 120.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9391,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T08:17:39.721Z",
          "last_at": "2026-09-15T08:17:39.721Z",
          "caps": [],
          "est_tokens_per_sec": 4.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.7484,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.5828,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37",
            "2026-W38"
          ],
          "tokens_per_sec_best": 189.43,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-09-13T17:53:41.023Z",
          "last_at": "2026-09-17T10:58:24.251Z",
          "caps": [
            "code.reason",
            "lang.en",
            "lang.es",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking"
          ],
          "usd_per_m_out": 1.3596,
          "usd_per_m_energy": 0.033,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q2_K",
      "family": "phi-4",
      "member": "mini-reasoning",
      "quant": "Q2_K",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-reasoning",
      "hub_repo": "microsoft/Phi-4-mini-reasoning",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 5,
      "disk_mib": 1088,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-reasoning",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/amd64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-27T21:18:14.816Z",
          "last_at": "2026-08-27T21:18:14.816Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 188.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3807,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7189,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1252.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.381,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 433.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0722,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 175.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6399,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 37.71,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-26T07:50:02.772Z",
          "last_at": "2026-08-26T07:50:02.772Z",
          "caps": [
            "code.reason",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.036,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.8373,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 9.83,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-30T05:16:19.603Z",
          "last_at": "2026-08-30T05:16:19.603Z",
          "caps": [
            "lang.en",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking"
          ],
          "est_usd_per_m_out": 4.2158,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 636.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4042,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q2_K_L",
      "family": "phi-4",
      "member": "mini-reasoning",
      "quant": "Q2_K_L",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-reasoning",
      "hub_repo": "microsoft/Phi-4-mini-reasoning",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 5,
      "disk_mib": 1088,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-reasoning",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W34",
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 188.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3807,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7189,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1252.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.381,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 433.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0722,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 175.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6399,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W34",
            "2026-W35"
          ],
          "tokens_per_sec_best": 6.82,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-22T04:28:06.206Z",
          "last_at": "2026-08-26T21:37:46.597Z",
          "caps": [
            "code.reason",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.036,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.8373,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 636.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4042,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q3_K_M",
      "family": "phi-4",
      "member": "mini-reasoning",
      "quant": "Q3_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-reasoning",
      "hub_repo": "microsoft/Phi-4-mini-reasoning",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 5,
      "disk_mib": 1088,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-reasoning",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W34",
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 188.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3807,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7189,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1252.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.381,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 433.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0722,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 175.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6399,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W34",
            "2026-W35"
          ],
          "tokens_per_sec_best": 31.94,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-23T17:13:58.854Z",
          "last_at": "2026-08-27T12:08:21.915Z",
          "caps": [
            "code.reason",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.036,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.8373,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 636.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4042,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Phi-4-mini-reasoning-GGUF:Q4_K_M",
      "family": "phi-4",
      "member": "mini-reasoning",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-reasoning",
      "hub_repo": "microsoft/Phi-4-mini-reasoning",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 2376,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-reasoning",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/amd64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 1.85,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-27T21:59:23.346Z",
          "last_at": "2026-08-27T21:59:23.346Z",
          "caps": [
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking"
          ],
          "est_usd_per_m_out": 22.9671,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 86.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8434,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6224,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 573.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.834,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 198.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.163,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 80.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4103,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 22.2274,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.9266,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 6.5,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-30T06:11:35.062Z",
          "last_at": "2026-08-30T06:11:35.062Z",
          "caps": [
            "code.reason",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking"
          ],
          "est_usd_per_m_out": 6.5368,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 291.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8862,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Phi-4-reasoning-plus-GGUF:Q4_K_M",
      "family": "phi-4",
      "member": "reasoning-plus",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "microsoft/Phi-4-reasoning-plus",
      "hub_repo": "microsoft/Phi-4-reasoning-plus",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 8633,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-reasoning-plus",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 1.6805
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 23.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.2786,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.8201,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 157.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.0624,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 22.28,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 6.48,
          "first_at": "2026-10-23T18:52:51.415Z",
          "last_at": "2026-10-23T18:52:51.415Z",
          "caps": [
            "code.reason",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "usd_per_m_out": 1.6805,
          "usd_per_m_energy": 0.18,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.3547,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 86.6642,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 50.3211,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 80.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.2836,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3-30B-A3B-Instruct-2507-GGUF:Q4_K_M",
      "family": "qwen-3",
      "member": "30b-a3b-instruct-2507",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-30B-A3B-Instruct-2507",
      "hub_repo": "Qwen/Qwen3-30B-A3B-Instruct-2507",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 17697,
      "active_share": 0.1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-30B-A3B-Instruct-2507",
        "fetched_at": "2026-11-10T15:57:15.731Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3-30b-a3b-instruct-2507",
        "ids": [
          "qwen/qwen3-30b-a3b-instruct-2507"
        ],
        "usd_per_m_in": 0.1,
        "usd_per_m_out": 0.3,
        "usd_per_m_cached_in": 0.1,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 116.05,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7357,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 26.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6742,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 770.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6373,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 266.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1681,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 108.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1659,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.4,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19.4032,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.2779,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 391.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.692,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3-4B-GGUF:BF16",
      "family": "qwen-3",
      "member": "4b",
      "quant": "BF16",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-4B",
      "hub_repo": "Qwen/Qwen3-4B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 4,
      "disk_mib": 7678,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-4B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W38"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "windows/amd64/nvidia/16g",
        "usd_per_m_out": 3.2504
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 26.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.8868,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.9425,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 177.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7193,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 61.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5969,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 24.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.7319,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 76.456,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 44.3797,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 80.76,
          "effective_ctx_best": 16384,
          "joules_per_token_best": null,
          "first_at": "2026-09-16T20:51:45.252Z",
          "last_at": "2026-09-16T20:51:45.252Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.ta",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "usd_per_m_out": 3.2504,
          "usd_per_m_energy": 0.0774,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ1_S",
      "family": "qwen-3",
      "member": "4b",
      "quant": "IQ1_S",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-4B",
      "hub_repo": "Qwen/Qwen3-4B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 1032,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-4B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W35",
        "2026-W38",
        "2026-W40",
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.5096
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W40"
          ],
          "tokens_per_sec_best": 14.37,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-01T06:21:07.060Z",
          "last_at": "2026-10-01T06:21:07.060Z",
          "caps": [],
          "usd_per_m_out": 4.9974,
          "usd_per_m_energy": 0.0677,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 45.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.681,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1320.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3614,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 61.33,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.975,
          "first_at": "2026-10-20T10:39:28.096Z",
          "last_at": "2026-10-20T10:39:28.096Z",
          "caps": [],
          "usd_per_m_out": 0.5096,
          "usd_per_m_energy": 0.0549,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 185.22,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6067,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 24.9,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-25T07:24:14.148Z",
          "last_at": "2026-08-25T07:24:14.148Z",
          "caps": [],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.5115,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.5325,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 90.41,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-14T03:37:24.095Z",
          "last_at": "2026-09-14T03:37:24.095Z",
          "caps": [],
          "usd_per_m_out": 2.8436,
          "usd_per_m_energy": 0.0691,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3-4B-GGUF:IQ3_XXS",
      "family": "qwen-3",
      "member": "4b",
      "quant": "IQ3_XXS",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3-4B",
      "hub_repo": "Qwen/Qwen3-4B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 2,
      "disk_mib": 1596,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-4B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W35",
        "2026-W37",
        "2026-W38",
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.6634
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 128.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5616,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 29.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0686,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 854.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5594,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 47.81,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 3.905,
          "first_at": "2026-10-20T10:56:06.004Z",
          "last_at": "2026-10-20T10:56:06.004Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 0.6634,
          "usd_per_m_energy": 0.1085,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 119.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9421,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 0.22,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-26T03:56:05.145Z",
          "last_at": "2026-08-26T03:56:05.145Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.ta",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.8096,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 4.67,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-11T19:27:11.854Z",
          "last_at": "2026-09-13T05:17:06.817Z",
          "caps": [
            "faithful_tool_result",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 15.4756,
          "usd_per_m_energy": 0.2082,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 4.87,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-11T11:19:12.369Z",
          "last_at": "2026-09-11T15:50:55.585Z",
          "caps": [
            "faithful_tool_result",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "est_usd_per_m_out": 8.5943,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 53.3,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-09-14T03:49:21.399Z",
          "last_at": "2026-09-14T03:49:21.399Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.ta",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 4.8321,
          "usd_per_m_energy": 0.1173,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF:Q4_K_M",
      "family": "qwen-3",
      "member": "coder-30b-a3b",
      "quant": "Q4_K_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 0.1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-Coder-30B-A3B-Instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3-coder-30b-a3b-instruct",
        "ids": [
          "qwen/qwen3-coder-30b-a3b-instruct"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 26.66,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 10.891,
          "first_at": "2026-10-22T20:20:24.815Z",
          "last_at": "2026-10-22T20:20:24.815Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3-VL-4B-Instruct-GGUF:IQ4_XS",
      "family": "qwen-3",
      "member": "vl-4b",
      "quant": "IQ4_XS",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "unsloth/Qwen3-VL-4B-Instruct-GGUF",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W34"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W34"
          ],
          "tokens_per_sec_best": 4.35,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-23T19:19:20.558Z",
          "last_at": "2026-08-23T19:19:20.558Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.nl",
            "lang.ru",
            "lang.ta",
            "lang.te",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "vision",
            "vision.count",
            "vision.read",
            "vision.spatial"
          ],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3-VL-4B-Instruct-GGUF:Q2_K",
      "family": "qwen-3",
      "member": "vl-4b",
      "quant": "Q2_K",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "unsloth/Qwen3-VL-4B-Instruct-GGUF",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W34"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W34"
          ],
          "tokens_per_sec_best": 11.79,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-22T10:25:26.258Z",
          "last_at": "2026-08-22T10:25:26.258Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop",
            "vision",
            "vision.count",
            "vision.read",
            "vision.spatial"
          ],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3-VL-4B-Instruct-GGUF:Q4_0",
      "family": "qwen-3",
      "member": "vl-4b",
      "quant": "Q4_0",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "unsloth/Qwen3-VL-4B-Instruct-GGUF",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 3,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 40.87,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-26T21:19:54.478Z",
          "last_at": "2026-08-27T18:43:47.183Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.read",
            "vision.spatial"
          ],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3-VL-4B-Instruct-GGUF:Q4_1",
      "family": "qwen-3",
      "member": "vl-4b",
      "quant": "Q4_1",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "unsloth/Qwen3-VL-4B-Instruct-GGUF",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W34",
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W34",
            "2026-W35"
          ],
          "tokens_per_sec_best": 37.02,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-23T23:33:49.043Z",
          "last_at": "2026-08-26T09:37:36.094Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.ml",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.read",
            "vision.spatial"
          ],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q4_0",
      "family": "qwen-3-5",
      "member": "0.8b",
      "quant": "Q4_0",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.5-0.8B",
      "hub_repo": "Qwen/Qwen3.5-0.8B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 3,
      "disk_mib": 483,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.5-0.8B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 1.8639
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 425.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1678,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 16.53,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.638,
          "first_at": "2026-10-25T21:02:42.313Z",
          "last_at": "2026-10-25T21:02:42.313Z",
          "caps": [
            "instruct.follow",
            "lang.en",
            "reason.channel",
            "thinking",
            "tools",
            "tools.call"
          ],
          "usd_per_m_out": 1.8639,
          "usd_per_m_energy": 0.0455,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2822.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.169,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 976.59,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0315,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 395.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2828,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.4244,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 27.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.5736,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1433.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1791,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q4_K_M",
      "family": "qwen-3-5",
      "member": "0.8b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.5-0.8B",
      "hub_repo": "Qwen/Qwen3.5-0.8B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 507,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.5-0.8B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 1.9149
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 405.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1762,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43",
            "2026-W44"
          ],
          "tokens_per_sec_best": 16.1,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T04:57:09.043Z",
          "last_at": "2026-10-27T03:30:04.973Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.9149,
          "usd_per_m_energy": 0.0259,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2688.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1774,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 930.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0331,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 377.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2969,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.37,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.6444,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 26.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7019,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1365.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.188,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3.5-0.8B-GGUF:Q8_0",
      "family": "qwen-3-5",
      "member": "0.8b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.5-0.8B",
      "hub_repo": "Qwen/Qwen3.5-0.8B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 1,
      "disk_mib": 774,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.5-0.8B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 2.5344
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 265.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2699,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 12.25,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.971,
          "first_at": "2026-10-25T21:33:46.195Z",
          "last_at": "2026-10-25T21:33:46.195Z",
          "caps": [
            "reason.channel",
            "thinking",
            "tools",
            "tools.call"
          ],
          "usd_per_m_out": 2.5344,
          "usd_per_m_energy": 0.0548,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1761.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2709,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 609.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0509,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 246.95,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4542,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.1104,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.1389,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 894.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2872,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3.5-27B-GGUF:Q4_K_M",
      "family": "qwen-3-5",
      "member": "27b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.5-27B",
      "hub_repo": "Qwen/Qwen3.5-27B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 15965,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.5-27B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.5-27b",
        "ids": [
          "qwen/qwen3.5-27b"
        ],
        "usd_per_m_in": 0.26,
        "usd_per_m_out": 2.6,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.5291,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.97,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.6158,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 85.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.7334,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 29.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.469,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.97,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.4027,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 171.3563,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 99.9578,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.2095,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_M",
      "family": "qwen-3-5",
      "member": "35b-a3b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.5-35B-A3B",
      "hub_repo": "Qwen/Qwen3.5-35B-A3B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 20996,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.5-35B-A3B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.5-35b-a3b",
        "ids": [
          "qwen/qwen3.5-35b-a3b"
        ],
        "usd_per_m_in": 0.14,
        "usd_per_m_out": 1,
        "usd_per_m_cached_in": 0.05,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 11.3644,
        "vs_api": 11.36
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 114.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7716,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W45"
          ],
          "tokens_per_sec_best": 4.18,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 9.026,
          "first_at": "2026-10-30T03:32:20.049Z",
          "last_at": "2026-11-04T07:52:33.352Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.es",
            "lang.ha",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ne",
            "lang.nl",
            "lang.pa",
            "lang.si",
            "lang.ta",
            "lang.te",
            "lang.ur",
            "output.long",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call"
          ],
          "usd_per_m_out": 11.3644,
          "usd_per_m_energy": 0.2507,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 757.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6517,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 262.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1812,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 106.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2109,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 20.3369,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.8359,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 384.63,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7106,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3.5-35B-A3B-GGUF:Q4_K_S",
      "family": "qwen-3-5",
      "member": "35b-a3b",
      "quant": "Q4_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.5-35B-A3B",
      "hub_repo": "Qwen/Qwen3.5-35B-A3B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 3,
      "disk_mib": 19716,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.5-35B-A3B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.5-35b-a3b",
        "ids": [
          "qwen/qwen3.5-35b-a3b"
        ],
        "usd_per_m_in": 0.14,
        "usd_per_m_out": 1,
        "usd_per_m_cached_in": 0.05,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 11.6738,
        "vs_api": 11.67
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 121.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.716,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 3.98,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 9.303,
          "first_at": "2026-11-04T13:28:00.581Z",
          "last_at": "2026-11-04T13:28:00.581Z",
          "caps": [
            "code.reason",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call"
          ],
          "usd_per_m_out": 11.6738,
          "usd_per_m_energy": 0.2584,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 806.63,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6107,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 279.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1665,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 113.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1279,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 18.8757,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.9732,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 409.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6648,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3.5-4B-GGUF:Q4_K_M",
      "family": "qwen-3-5",
      "member": "4b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.5-4B",
      "hub_repo": "Qwen/Qwen3.5-4B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 2613,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.5-4B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 2.7177
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 78.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.93,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 18.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.795,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 521.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9175,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 180.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1803,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 41.82,
          "effective_ctx_best": 262144,
          "joules_per_token_best": 1.836,
          "first_at": "2026-11-07T00:13:54.843Z",
          "last_at": "2026-11-07T00:13:54.843Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 2.7177,
          "usd_per_m_energy": 0.051,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 24.5298,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.2493,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 264.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9754,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3.5-9B-GGUF:Q4_K_M",
      "family": "qwen-3-5",
      "member": "9b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.5-9B",
      "hub_repo": "Qwen/Qwen3.5-9B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 5417,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.5-9B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.5-9b",
        "ids": [
          "qwen/qwen3.5-9b"
        ],
        "usd_per_m_in": 0.037,
        "usd_per_m_out": 0.054,
        "usd_per_m_cached_in": 0.033,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 37.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.9884,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.9846,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 251.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.9112,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 87.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3999,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 35.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.2852,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 52.3476,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 30.5185,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 127.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.0399,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3.6-27B-GGUF:Q4_K_M",
      "family": "qwen-3-6",
      "member": "27b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.6-27B",
      "hub_repo": "Qwen/Qwen3.6-27B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 16038,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.6-27B",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.6-27b",
        "ids": [
          "qwen/qwen3.6-27b"
        ],
        "usd_per_m_in": 0.32,
        "usd_per_m_out": 3.2,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 22.5279,
        "vs_api": 7.04
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.81,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.5592,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.95,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.7351,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.7598,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 29.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.478,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 5.53,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 14.967,
          "first_at": "2026-11-06T21:35:21.155Z",
          "last_at": "2026-11-06T21:35:21.155Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 22.5279,
          "usd_per_m_energy": 0.4158,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 171.4775,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 100.0285,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.2396,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3.6-35B-A3B-GGUF:Q4_K_M",
      "family": "qwen-3-6",
      "member": "35b-a3b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.6-35B-A3B",
      "hub_repo": "Qwen/Qwen3.6-35B-A3B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 21109,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.6-35B-A3B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.6-35b-a3b",
        "ids": [
          "qwen/qwen3.6-35b-a3b"
        ],
        "usd_per_m_in": 0.1,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.025,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 113.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7766,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 26.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.818,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 753.4,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6553,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 260.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1826,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 105.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2183,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 20.4526,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.4,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.9123,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 382.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7147,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3.6-35B-A3B-MTP-GGUF:Q4_K_M",
      "family": "qwen-3-6",
      "member": "35b-a3b~unsloth-mtp",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.6-35B-A3B",
      "hub_repo": "Qwen/Qwen3.6-35B-A3B",
      "publisher": "unsloth",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 21613,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.6-35B-A3B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.6-35b-a3b",
        "ids": [
          "qwen/qwen3.6-35b-a3b"
        ],
        "usd_per_m_in": 0.1,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.025,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 110.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7989,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 25.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8774,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 735.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6715,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 254.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1885,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 103.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2514,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 21.0358,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.2491,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 373.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7329,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3.8-27B-GGUF:Q4_K_M",
      "family": "qwen-3-8",
      "member": "27b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 15796,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.4482,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.4238,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 86.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.6707,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 29.86,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4491,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.2796,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 171.0756,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 98.62,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.1411,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/Qwen3.8-27B-GGUF:Q8_0",
      "family": "qwen-3-8",
      "member": "27b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 1,
      "disk_mib": 27701,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 19.1191,
        "vs_api": 10.2
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.6202,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 30.9706,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 49.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.1417,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W45"
          ],
          "tokens_per_sec_best": 2.77,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 202.09,
          "first_at": "2026-10-28T06:43:31.071Z",
          "last_at": "2026-11-03T18:50:36.027Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.ha",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "lang.yo",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 19.1191,
          "usd_per_m_energy": 5.6136,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 19.4306,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 333.9834,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 194.8237,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 24.99,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.1562,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/embeddinggemma-300m-GGUF:BF16",
      "family": "embeddinggemma",
      "member": "300m",
      "quant": "BF16",
      "engine": "ollama",
      "modality": "text",
      "base": "google/embeddinggemma-300m",
      "hub_repo": "google/embeddinggemma-300m",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 4,
      "disk_mib": 584,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/embeddinggemma-300m",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 351.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2032,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:33:53.298Z",
          "last_at": "2026-10-25T20:33:53.298Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 81.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3809,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2334.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2044,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 807.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0382,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 327.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3422,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.3559,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 22.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.1159,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1185.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2166,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/embeddinggemma-300m-GGUF:F32",
      "family": "embeddinggemma",
      "member": "300m",
      "quant": "F32",
      "engine": "ollama",
      "modality": "text",
      "base": "google/embeddinggemma-300m",
      "hub_repo": "google/embeddinggemma-300m",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 1161,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/embeddinggemma-300m",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 176.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4066,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T23:47:58.918Z",
          "last_at": "2026-10-25T23:47:58.918Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 40.8,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7687,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T02:37:19.884Z",
          "last_at": "2026-11-03T02:37:19.884Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 1174.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4067,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 406.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0772,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 164.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6831,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.7179,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 11.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.232,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 596.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4314,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/embeddinggemma-300m-GGUF:Q4_0",
      "family": "embeddinggemma",
      "member": "300m",
      "quant": "Q4_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/embeddinggemma-300m",
      "hub_repo": "google/embeddinggemma-300m",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 3,
      "disk_mib": 264,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/embeddinggemma-300m",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 777.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0915,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:32:49.448Z",
          "last_at": "2026-10-25T20:32:49.448Z",
          "caps": [
            "embed.vector"
          ],
          "est_tokens_per_sec": 179.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1707,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5163.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0923,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1786.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0171,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 724.03,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1543,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 29.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.4123,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 50.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4033,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2621.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0978,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:BF16",
      "family": "functiongemma",
      "member": "270m",
      "quant": "BF16",
      "engine": "ollama",
      "modality": "text",
      "base": "google/functiongemma-270m-it",
      "hub_repo": "google/functiongemma-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 4,
      "disk_mib": 517,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/functiongemma-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 1.6482
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 397.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1797,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 18.71,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.447,
          "first_at": "2026-10-25T01:54:32.256Z",
          "last_at": "2026-10-25T01:54:32.256Z",
          "caps": [
            "lang.bn",
            "lang.en",
            "lang.hi",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.6482,
          "usd_per_m_energy": 0.068,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2636.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1809,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 912.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0338,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 369.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3028,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.7374,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 25.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7554,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1338.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1917,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:IQ2_M",
      "family": "functiongemma",
      "member": "270m",
      "quant": "IQ2_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/functiongemma-270m-it",
      "hub_repo": "google/functiongemma-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 173,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/functiongemma-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9402
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1187.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0599,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 32.5,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.115,
          "first_at": "2026-10-25T01:21:02.968Z",
          "last_at": "2026-10-25T01:21:02.968Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.9402,
          "usd_per_m_energy": 0.031,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7879.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0605,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2726.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0112,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1104.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1011,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 45.03,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.5792,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 77.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9185,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4001.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0641,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:IQ2_XXS",
      "family": "functiongemma",
      "member": "270m",
      "quant": "IQ2_XXS",
      "engine": "ollama",
      "modality": "text",
      "base": "google/functiongemma-270m-it",
      "hub_repo": "google/functiongemma-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 171,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/functiongemma-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "windows/amd64/intel/16g",
        "usd_per_m_out": 1.0652
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1201.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0592,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 277.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1103,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7971.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0598,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2758.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0111,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1117.8,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0999,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 45.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.5608,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": 66.76,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.892,
          "first_at": "2026-11-14T04:16:02.312Z",
          "last_at": "2026-11-14T04:16:02.312Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.0652,
          "usd_per_m_energy": 0.0526,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4047.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0633,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:IQ3_XXS",
      "family": "functiongemma",
      "member": "270m",
      "quant": "IQ3_XXS",
      "engine": "ollama",
      "modality": "text",
      "base": "google/functiongemma-270m-it",
      "hub_repo": "google/functiongemma-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 2,
      "disk_mib": 175,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/functiongemma-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9365
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1173.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0606,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 32.63,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.446,
          "first_at": "2026-10-25T01:27:13.244Z",
          "last_at": "2026-10-25T01:27:13.244Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.9365,
          "usd_per_m_energy": 0.0679,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7789.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0612,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2695.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0113,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1092.25,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1022,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 44.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.5974,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 76.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9291,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3955.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0648,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_0",
      "family": "functiongemma",
      "member": "270m",
      "quant": "Q4_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/functiongemma-270m-it",
      "hub_repo": "google/functiongemma-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 3,
      "disk_mib": 230,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/functiongemma-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.8216
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 892.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0797,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 37.25,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.299,
          "first_at": "2026-10-25T01:30:48.038Z",
          "last_at": "2026-10-25T01:30:48.038Z",
          "caps": [
            "code.reason",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.8216,
          "usd_per_m_energy": 0.0361,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5926.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0804,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2050.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0149,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 831.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1344,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 33.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.101,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 58.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.222,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3009.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0852,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q4_K_M",
      "family": "functiongemma",
      "member": "270m",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/functiongemma-270m-it",
      "hub_repo": "google/functiongemma-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 241,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/functiongemma-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 0.3278
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 852.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0835,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 33.83,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.674,
          "first_at": "2026-10-24T21:49:19.289Z",
          "last_at": "2026-10-24T21:49:19.289Z",
          "caps": [
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.9049,
          "usd_per_m_energy": 0.0465,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5656.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0843,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1957.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0156,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 340.82,
          "effective_ctx_best": null,
          "joules_per_token_best": 6.748,
          "first_at": "2026-11-06T22:25:06.253Z",
          "last_at": "2026-11-06T22:25:06.253Z",
          "caps": [
            "lang.en",
            "lang.hi",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.3278,
          "usd_per_m_energy": 0.1874,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 32.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.2013,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": 23.66,
          "effective_ctx_best": null,
          "joules_per_token_best": 9.87,
          "first_at": "2026-11-10T22:17:22.827Z",
          "last_at": "2026-11-10T22:17:22.827Z",
          "caps": [
            "code.write",
            "lang.bn",
            "lang.hi",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 3.008,
          "usd_per_m_energy": 0.2742,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2872.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0893,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/functiongemma-270m-it-GGUF:Q8_0",
      "family": "functiongemma",
      "member": "270m",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/functiongemma-270m-it",
      "hub_repo": "google/functiongemma-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 1,
      "disk_mib": 278,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/functiongemma-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9314
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 738.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0964,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 32.9,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.99,
          "first_at": "2026-10-25T01:42:40.647Z",
          "last_at": "2026-10-25T01:42:40.647Z",
          "caps": [
            "lang.en",
            "lang.hi",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.9314,
          "usd_per_m_energy": 0.0553,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4903.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0972,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1696.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0181,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 687.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1625,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 28.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.541,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 48.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4778,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2489.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.103,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:F16",
      "family": "gemma-3",
      "member": "270m",
      "quant": "F16",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-3-270m-it",
      "hub_repo": "google/gemma-3-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 4,
      "disk_mib": 517,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 1.3111
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 397.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1797,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 23.52,
          "effective_ctx_best": 16384,
          "joules_per_token_best": 1.245,
          "first_at": "2026-10-25T01:10:04.264Z",
          "last_at": "2026-10-25T01:10:04.264Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.3111,
          "usd_per_m_energy": 0.0346,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2636.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1809,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 912.36,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0338,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 369.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3028,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.07,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.7374,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 25.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7554,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1338.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1917,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_M",
      "family": "gemma-3",
      "member": "270m",
      "quant": "IQ2_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-3-270m-it",
      "hub_repo": "google/gemma-3-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 174,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.8837
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1180.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0603,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 34.58,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.168,
          "first_at": "2026-10-25T00:21:36.472Z",
          "last_at": "2026-10-25T00:21:36.472Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.8837,
          "usd_per_m_energy": 0.0324,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7834.22,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0608,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2710.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0113,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1098.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1017,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 44.77,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.5884,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": 68.95,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.382,
          "first_at": "2026-11-14T04:18:28.711Z",
          "last_at": "2026-11-14T04:18:28.711Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.0314,
          "usd_per_m_energy": 0.0384,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3978.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0645,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ2_XXS",
      "family": "gemma-3",
      "member": "270m",
      "quant": "IQ2_XXS",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-3-270m-it",
      "hub_repo": "google/gemma-3-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 171,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.8862
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1201.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0592,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 34.48,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.58,
          "first_at": "2026-10-25T00:18:03.720Z",
          "last_at": "2026-10-25T00:18:03.720Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.8862,
          "usd_per_m_energy": 0.0439,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7971.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0598,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2758.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0111,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1117.8,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0999,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 45.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.5608,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": 34.6,
          "effective_ctx_best": null,
          "joules_per_token_best": 3.406,
          "first_at": "2026-11-14T02:40:41.929Z",
          "last_at": "2026-11-14T02:40:41.929Z",
          "caps": [
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 2.0552,
          "usd_per_m_energy": 0.0946,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4047.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0633,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:IQ3_XXS",
      "family": "gemma-3",
      "member": "270m",
      "quant": "IQ3_XXS",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-3-270m-it",
      "hub_repo": "google/gemma-3-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 2,
      "disk_mib": 175,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9147
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1173.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0606,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 33.41,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.92,
          "first_at": "2026-10-25T00:25:01.566Z",
          "last_at": "2026-10-25T00:25:01.566Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.9147,
          "usd_per_m_energy": 0.0533,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7789.45,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0612,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2695.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0113,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1092.25,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1022,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 44.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.5974,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": 52.33,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.809,
          "first_at": "2026-11-14T17:02:29.684Z",
          "last_at": "2026-11-14T17:02:29.684Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.359,
          "usd_per_m_energy": 0.078,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3955.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0648,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_0",
      "family": "gemma-3",
      "member": "270m",
      "quant": "Q4_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-3-270m-it",
      "hub_repo": "google/gemma-3-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 3,
      "disk_mib": 230,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.9274
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 892.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0797,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 33,
          "effective_ctx_best": null,
          "joules_per_token_best": 0.854,
          "first_at": "2026-10-25T00:28:50.582Z",
          "last_at": "2026-10-25T00:28:50.582Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.9274,
          "usd_per_m_energy": 0.0237,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5926.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0804,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2050.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0149,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 831.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1344,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 33.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.101,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 58.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.222,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3009.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0852,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q4_K_M",
      "family": "gemma-3",
      "member": "270m",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-3-270m-it",
      "hub_repo": "google/gemma-3-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 241,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 0.3829
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 852.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0835,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 37.38,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.38,
          "first_at": "2026-10-24T21:43:10.768Z",
          "last_at": "2026-10-24T21:43:10.768Z",
          "caps": [
            "code.write",
            "lang.bn",
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.819,
          "usd_per_m_energy": 0.0383,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5656.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0843,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1957.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0156,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 291.82,
          "effective_ctx_best": null,
          "joules_per_token_best": 3.229,
          "first_at": "2026-11-06T10:24:40.004Z",
          "last_at": "2026-11-06T10:24:40.004Z",
          "caps": [
            "code.reason",
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.3829,
          "usd_per_m_energy": 0.0897,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 27.34,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.5,
          "first_at": "2026-10-28T13:51:02.371Z",
          "last_at": "2026-10-28T13:51:02.371Z",
          "caps": [
            "code.reason",
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 2.6031,
          "usd_per_m_energy": 0.0417,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W46"
          ],
          "tokens_per_sec_best": 51.51,
          "effective_ctx_best": null,
          "joules_per_token_best": 2.824,
          "first_at": "2026-10-26T04:40:08.418Z",
          "last_at": "2026-11-11T05:29:13.776Z",
          "caps": [
            "code.write",
            "lang.en",
            "lang.ta",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 1.3816,
          "usd_per_m_energy": 0.0784,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 286.08,
          "effective_ctx_best": null,
          "joules_per_token_best": 12.384,
          "first_at": "2026-11-08T15:41:20.899Z",
          "last_at": "2026-11-08T15:41:20.899Z",
          "caps": [
            "code.reason",
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.8964,
          "usd_per_m_energy": 0.344,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-3-270m-it-GGUF:Q8_0",
      "family": "gemma-3",
      "member": "270m",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-3-270m-it",
      "hub_repo": "google/gemma-3-270m-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 1,
      "disk_mib": 278,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.993
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 738.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0964,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 30.86,
          "effective_ctx_best": 32768,
          "joules_per_token_best": 1.895,
          "first_at": "2026-10-25T00:53:07.095Z",
          "last_at": "2026-10-25T00:53:07.095Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.schema",
            "structured_output"
          ],
          "usd_per_m_out": 0.993,
          "usd_per_m_energy": 0.0526,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4903.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0972,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1696.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0181,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 687.56,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1625,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 28.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.541,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 48.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4778,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2489.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.103,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-3-27b-it-GGUF:Q4_K_M",
      "family": "gemma-3",
      "member": "27b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-3-27b-it",
      "hub_repo": "google/gemma-3-27b-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 15780,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-27b-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "google/gemma-3-27b-it",
        "ids": [
          "google/gemma-3-27b-it"
        ],
        "usd_per_m_in": 0.08,
        "usd_per_m_out": 0.35,
        "usd_per_m_cached_in": 0.04,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.4423,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 14.4195,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 86.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.6653,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 29.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4473,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.27,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 171.049,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 98.6047,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.1338,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-3-4b-it-GGUF:Q4_K_M",
      "family": "gemma-3",
      "member": "4b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-3-4b-it",
      "hub_repo": "google/gemma-3-4b-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 2374,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-4b-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "google/gemma-3-4b-it",
        "ids": [
          "google/gemma-3-4b-it"
        ],
        "usd_per_m_in": 0.05,
        "usd_per_m_out": 0.1,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 86.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8427,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.95,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6215,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 574.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8333,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 198.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1628,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 80.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4091,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 22.2269,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.9263,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 291.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8855,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-12b-it-GGUF:BF16",
      "family": "gemma-4",
      "member": "12b",
      "quant": "BF16",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-12B-it",
      "hub_repo": "google/gemma-4-12B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 4,
      "disk_mib": 22728,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-12B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 4.5628
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.8969,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.08,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 23.5158,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 59.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.2531,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 10.72,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 14.417,
          "first_at": "2026-10-23T04:54:34.480Z",
          "last_at": "2026-10-24T08:48:51.145Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 4.5628,
          "usd_per_m_energy": 0.4005,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.41,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 15.4606,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 263.1421,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.59,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 151.6412,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 30.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.0199,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-12b-it-GGUF:IQ4_XS",
      "family": "gemma-4",
      "member": "12b",
      "quant": "IQ4_XS",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-12B-it",
      "hub_repo": "google/gemma-4-12B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 9,
      "disk_mib": 6080,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-12B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 1.5329
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 33.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.2475,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 7.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.5397,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 224.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.1475,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 23.07,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 4.775,
          "first_at": "2026-10-23T15:47:04.536Z",
          "last_at": "2026-10-23T15:47:04.536Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.5329,
          "usd_per_m_energy": 0.1326,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 31.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.7047,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 59.3126,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 34.5092,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 113.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.2942,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-12b-it-GGUF:Q4_K_M",
      "family": "gemma-4",
      "member": "12b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-12B-it",
      "hub_repo": "google/gemma-4-12B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 6791,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-12B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 1.3594
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 30.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.5297,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.1495,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 200.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.4015,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 26.44,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 5.962,
          "first_at": "2026-10-23T01:18:08.196Z",
          "last_at": "2026-10-24T16:11:48.110Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.ha",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.3594,
          "usd_per_m_energy": 0.1656,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 28.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.1582,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 66.5207,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.97,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 38.8319,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 101.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.5682,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-12b-it-GGUF:Q8_0",
      "family": "gemma-4",
      "member": "12b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-12B-it",
      "hub_repo": "google/gemma-4-12B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 1,
      "disk_mib": 12082,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-12B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 2.2066
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.7532,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.92,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.2676,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 112.83,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.3106,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 18.24,
          "effective_ctx_best": 262144,
          "joules_per_token_best": 8.615,
          "first_at": "2026-10-23T13:42:31.041Z",
          "last_at": "2026-10-24T17:37:21.280Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 2.2066,
          "usd_per_m_energy": 0.2393,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.6713,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 126.2572,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 72.7969,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 57.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.6445,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:Q4_K_M",
      "family": "gemma-4",
      "member": "26b-a4b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-26B-A4B-it",
      "hub_repo": "google/gemma-4-26B-A4B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 16162,
      "active_share": 0.1538,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-26B-A4B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "google/gemma-4-26b-a4b-it",
        "ids": [
          "google/gemma-4-26b-a4b-it",
          "google/gemma-4-26b-a4b-it:free"
        ],
        "usd_per_m_in": 0.0675,
        "usd_per_m_out": 0.225,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 262144,
        "free_variant": true,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 82.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0185,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.05,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.2871,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 548.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8932,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 189.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2297,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 76.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.622,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.877,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 15.6076,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 278.39,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9677,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-26B-A4B-it-GGUF:Q8_0",
      "family": "gemma-4",
      "member": "26b-a4b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-26B-A4B-it",
      "hub_repo": "google/gemma-4-26B-A4B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 1,
      "disk_mib": 25615,
      "active_share": 0.1538,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-26B-A4B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "google/gemma-4-26b-a4b-it",
        "ids": [
          "google/gemma-4-26b-a4b-it",
          "google/gemma-4-26b-a4b-it:free"
        ],
        "usd_per_m_in": 0.0675,
        "usd_per_m_out": 0.225,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 262144,
        "free_variant": true,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 2.4693,
        "vs_api": 10.97
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 52.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7617,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 12.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.2647,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 345.91,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4379,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 20.76,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 6.525,
          "first_at": "2026-10-23T05:37:58.566Z",
          "last_at": "2026-10-24T13:57:16.814Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.es",
            "lang.gu",
            "lang.ha",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ne",
            "lang.nl",
            "lang.or",
            "lang.pa",
            "lang.ru",
            "lang.si",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "lang.ur",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 2.4693,
          "usd_per_m_energy": 0.1812,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 48.5,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7293,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 46.3726,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.4,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 27.0052,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 175.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.5775,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-31B-it-GGUF:Q4_0",
      "family": "gemma-4",
      "member": "31b",
      "quant": "Q4_0",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "unsloth/gemma-4-31B-it-GGUF",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-31B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "google/gemma-4-31b-it",
        "ids": [
          "google/gemma-4-31b-it",
          "google/gemma-4-31b-it:free"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 0.4,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": true,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 2.44,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 43.803,
          "first_at": "2026-10-23T04:09:16.960Z",
          "last_at": "2026-10-23T04:09:16.960Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-31B-it-GGUF:Q4_K_M",
      "family": "gemma-4",
      "member": "31b",
      "quant": "Q4_K_M",
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "unsloth/gemma-4-31B-it-GGUF",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-31B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": {
        "openrouter_id": "google/gemma-4-31b-it",
        "ids": [
          "google/gemma-4-31b-it",
          "google/gemma-4-31b-it:free"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 0.4,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": true,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 3.36,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 20.346,
          "first_at": "2026-10-22T21:50:45.315Z",
          "last_at": "2026-10-24T12:57:51.781Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.es",
            "lang.gu",
            "lang.ha",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ne",
            "lang.nl",
            "lang.or",
            "lang.pa",
            "lang.ru",
            "lang.si",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "lang.ur",
            "lang.yo",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:BF16",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "BF16",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E2B-it",
      "hub_repo": "google/gemma-4-E2B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 4,
      "disk_mib": 8879,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E2B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 1.2011
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 23.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.3808,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 3.14,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 10.27,
          "first_at": "2026-10-27T02:09:25.950Z",
          "last_at": "2026-10-27T02:09:25.950Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 11.988,
          "usd_per_m_energy": 0.2853,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 153.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.1509,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 31.34,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 3.472,
          "first_at": "2026-10-21T23:19:54.097Z",
          "last_at": "2026-10-24T14:39:11.148Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.es",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ne",
            "lang.nl",
            "lang.or",
            "lang.pa",
            "lang.ru",
            "lang.si",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "lang.ur",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.2011,
          "usd_per_m_energy": 0.0964,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 21.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.5157,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 88.8613,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 51.7867,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 77.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.3796,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_NL",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "IQ4_NL",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E2B-it",
      "hub_repo": "google/gemma-4-E2B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 2184,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E2B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W34",
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 94.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7736,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 21.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4843,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 624.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7663,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 215.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1491,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 87.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2946,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W34",
            "2026-W35"
          ],
          "tokens_per_sec_best": 54.91,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-22T07:03:32.829Z",
          "last_at": "2026-08-27T04:45:39.008Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 20.378,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.8678,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 316.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8141,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:IQ4_XS",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "IQ4_XS",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E2B-it",
      "hub_repo": "google/gemma-4-E2B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 2845,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E2B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W38",
        "2026-W43",
        "2026-W44",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.4088
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 72.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.0152,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 6.93,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 4.829,
          "first_at": "2026-10-27T04:51:25.171Z",
          "last_at": "2026-10-27T04:51:25.171Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 4.7232,
          "usd_per_m_energy": 0.1341,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 479.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9994,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43",
            "2026-W45"
          ],
          "tokens_per_sec_best": 80.07,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.267,
          "first_at": "2026-10-20T11:27:40.522Z",
          "last_at": "2026-11-03T04:32:02.604Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.es",
            "lang.gu",
            "lang.he",
            "lang.hi",
            "lang.id",
            "lang.ja",
            "lang.kn",
            "lang.ko",
            "lang.ml",
            "lang.mr",
            "lang.ms",
            "lang.ne",
            "lang.nl",
            "lang.or",
            "lang.pa",
            "lang.pl",
            "lang.ru",
            "lang.si",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "lang.th",
            "lang.ur",
            "lang.vi",
            "lang.zh",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.4088,
          "usd_per_m_energy": 0.0352,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 67.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6943,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.74,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.7473,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 15.56,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38",
            "2026-W45"
          ],
          "tokens_per_sec_best": 182.88,
          "effective_ctx_best": 8192,
          "joules_per_token_best": null,
          "first_at": "2026-09-14T06:26:26.969Z",
          "last_at": "2026-11-08T13:22:33.389Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.4139,
          "usd_per_m_energy": 0.0342,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q3_K_M",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "Q3_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E2B-it",
      "hub_repo": "google/gemma-4-E2B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 2184,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E2B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W34",
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/amd64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 2.78,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-28T08:20:27.355Z",
          "last_at": "2026-08-28T08:20:27.355Z",
          "caps": [
            "faithful_tool_result",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "est_usd_per_m_out": 15.2277,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 94.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7736,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 21.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4843,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 624.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7663,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 215.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1491,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 87.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2946,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W34",
            "2026-W35"
          ],
          "tokens_per_sec_best": 38.36,
          "effective_ctx_best": 8192,
          "joules_per_token_best": null,
          "first_at": "2026-08-21T12:57:15.347Z",
          "last_at": "2026-08-26T04:09:34.202Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 20.378,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.8678,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 316.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8141,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q3_K_S",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "Q3_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E2B-it",
      "hub_repo": "google/gemma-4-E2B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 2184,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E2B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W36",
        "2026-W37"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "windows/amd64/intel/16g",
        "usd_per_m_out": 5.5961
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 94.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7736,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 21.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4843,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 624.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7663,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 215.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1491,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 87.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2946,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 20.378,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": 13,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-09-12T14:34:10.214Z",
          "last_at": "2026-09-12T14:34:10.214Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.nl",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 5.5961,
          "usd_per_m_energy": 0.0748,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W36"
          ],
          "tokens_per_sec_best": 14.42,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-09-05T13:57:24.911Z",
          "last_at": "2026-09-05T13:57:24.911Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "est_usd_per_m_out": 2.9357,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 316.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8141,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_0",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "Q4_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E2B-it",
      "hub_repo": "google/gemma-4-E2B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 2184,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E2B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W35",
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.3806
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 94.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7736,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "darwin/arm64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 57.55,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-28T09:16:34.509Z",
          "last_at": "2026-08-28T09:16:34.509Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "est_usd_per_m_out": 0.7356,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 21.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.4843,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 624.15,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7663,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 84.59,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.244,
          "first_at": "2026-10-20T11:35:19.976Z",
          "last_at": "2026-10-20T11:35:19.976Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.3806,
          "usd_per_m_energy": 0.0346,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 87.52,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2946,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 20.378,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 11.8678,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 9.1,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-29T23:18:40.629Z",
          "last_at": "2026-08-29T23:18:40.629Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "est_usd_per_m_out": 4.652,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 316.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8141,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q4_K_M",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E2B-it",
      "hub_repo": "google/gemma-4-E2B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 3902,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E2B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W34",
        "2026-W37",
        "2026-W38",
        "2026-W40",
        "2026-W41",
        "2026-W42",
        "2026-W43",
        "2026-W44",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.433
      },
      "classes": {
        "darwin/amd64/apple/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 3.99,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-22T10:43:12.498Z",
          "last_at": "2026-10-22T10:43:12.498Z",
          "caps": [
            "faithful_tool_result",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 8.4191,
          "usd_per_m_energy": 0.1044,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 53.69,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 3.761,
          "first_at": "2026-11-02T09:23:57.166Z",
          "last_at": "2026-11-02T09:23:57.166Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.381,
          "usd_per_m_energy": 0.1045,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 24.89,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 3.606,
          "first_at": "2026-10-27T00:42:02.167Z",
          "last_at": "2026-10-27T00:42:02.167Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.3496,
          "usd_per_m_energy": 0.1002,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 291.17,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 2.228,
          "first_at": "2026-11-03T02:29:57.083Z",
          "last_at": "2026-11-03T02:29:57.083Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.6475,
          "usd_per_m_energy": 0.0619,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W43",
            "2026-W44",
            "2026-W45"
          ],
          "tokens_per_sec_best": 77.58,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 4.084,
          "first_at": "2026-10-19T11:17:50.473Z",
          "last_at": "2026-11-02T21:55:45.037Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.ar",
            "lang.as",
            "lang.bn",
            "lang.de",
            "lang.en",
            "lang.es",
            "lang.fa",
            "lang.fr",
            "lang.gu",
            "lang.he",
            "lang.hi",
            "lang.id",
            "lang.it",
            "lang.ja",
            "lang.kn",
            "lang.ko",
            "lang.ml",
            "lang.mr",
            "lang.ms",
            "lang.ne",
            "lang.nl",
            "lang.or",
            "lang.pa",
            "lang.pl",
            "lang.pt",
            "lang.ru",
            "lang.si",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "lang.th",
            "lang.tr",
            "lang.uk",
            "lang.ur",
            "lang.vi",
            "lang.zh",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.433,
          "usd_per_m_energy": 0.1134,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 98.85,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.866,
          "first_at": "2026-11-03T19:11:02.138Z",
          "last_at": "2026-11-06T05:45:12.775Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.1604,
          "usd_per_m_energy": 0.0518,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W34"
          ],
          "tokens_per_sec_best": 6.8,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-21T02:36:57.450Z",
          "last_at": "2026-08-21T02:36:57.450Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.nl",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 37.0738,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 6.8,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-14T02:26:48.645Z",
          "last_at": "2026-09-14T02:26:48.645Z",
          "caps": [
            "faithful_tool_result",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 10.9041,
          "usd_per_m_energy": 0.143,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W37",
            "2026-W38",
            "2026-W40",
            "2026-W41",
            "2026-W42"
          ],
          "tokens_per_sec_best": 97.95,
          "effective_ctx_best": 131072,
          "joules_per_token_best": null,
          "first_at": "2026-09-11T20:08:30.023Z",
          "last_at": "2026-10-12T14:30:23.535Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 2.6485,
          "usd_per_m_energy": 0.0638,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/8g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 15.33,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-16T10:35:09.210Z",
          "last_at": "2026-09-16T11:18:57.277Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "est_usd_per_m_out": 9.6476,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:Q8_0",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E2B-it",
      "hub_repo": "google/gemma-4-E2B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 1,
      "disk_mib": 4814,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E2B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.572
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 42.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7555,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 5.33,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 7.213,
          "first_at": "2026-10-27T11:32:11.254Z",
          "last_at": "2026-10-27T11:32:11.254Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 6.4417,
          "usd_per_m_energy": 0.2004,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 283.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6967,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 60.02,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.619,
          "first_at": "2026-10-21T08:31:23.897Z",
          "last_at": "2026-10-24T18:02:16.226Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.es",
            "lang.gu",
            "lang.hi",
            "lang.ja",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ne",
            "lang.nl",
            "lang.or",
            "lang.pa",
            "lang.ru",
            "lang.si",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "lang.ur",
            "lang.zh",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.572,
          "usd_per_m_energy": 0.045,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 39.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.9072,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 46.2283,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.9388,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 143.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.8094,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E2B-it-GGUF:q4_k_m",
      "family": "gemma-4",
      "member": "e2b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E2B-it",
      "hub_repo": "google/gemma-4-E2B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 3902,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E2B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W38",
        "2026-W40",
        "2026-W42",
        "2026-W43",
        "2026-W44",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.433
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 53.69,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 3.761,
          "first_at": "2026-11-02T09:23:57.166Z",
          "last_at": "2026-11-02T09:23:57.166Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.381,
          "usd_per_m_energy": 0.1045,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": 24.89,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 4.337,
          "first_at": "2026-10-27T00:42:02.167Z",
          "last_at": "2026-10-27T00:42:02.167Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.3496,
          "usd_per_m_energy": 0.1205,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 291.17,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 2.228,
          "first_at": "2026-11-03T02:29:57.083Z",
          "last_at": "2026-11-03T02:29:57.083Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.6475,
          "usd_per_m_energy": 0.0619,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W43",
            "2026-W45"
          ],
          "tokens_per_sec_best": 77.58,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 4.084,
          "first_at": "2026-10-19T11:17:50.473Z",
          "last_at": "2026-11-02T21:55:45.037Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.ja",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.nl",
            "lang.pa",
            "lang.pl",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.433,
          "usd_per_m_energy": 0.1134,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 98.85,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 1.866,
          "first_at": "2026-11-06T05:45:12.775Z",
          "last_at": "2026-11-06T05:45:12.775Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.nl",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.1604,
          "usd_per_m_energy": 0.0518,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 37.0738,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 6.8,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-14T02:26:48.645Z",
          "last_at": "2026-09-14T02:26:48.645Z",
          "caps": [
            "faithful_tool_result",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 10.9041,
          "usd_per_m_energy": 0.143,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W40",
            "2026-W42"
          ],
          "tokens_per_sec_best": 87.36,
          "effective_ctx_best": 131072,
          "joules_per_token_best": null,
          "first_at": "2026-09-30T17:20:42.216Z",
          "last_at": "2026-10-12T14:30:23.535Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 2.9696,
          "usd_per_m_energy": 0.0715,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/nvidia/8g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": 15.33,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-16T10:35:09.210Z",
          "last_at": "2026-09-16T11:18:57.277Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "est_usd_per_m_out": 9.6476,
          "est_price_basis": "the network's metered rates for a 4-thread machine of this class",
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E4B-it-GGUF:BF16",
      "family": "gemma-4",
      "member": "e4b",
      "quant": "BF16",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E4B-it",
      "hub_repo": "google/gemma-4-E4B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 4,
      "disk_mib": 14355,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E4B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 1.6336
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 14.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.776,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.7572,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 94.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.1412,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 25.77,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 6.529,
          "first_at": "2026-10-23T13:58:27.796Z",
          "last_at": "2026-10-24T15:19:31.012Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 1.6336,
          "usd_per_m_energy": 0.1814,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 13.32,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 9.25,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 153.0636,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.93,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 88.8756,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 48.22,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.5565,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E4B-it-GGUF:IQ4_XS",
      "family": "gemma-4",
      "member": "e4b",
      "quant": "IQ4_XS",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E4B-it",
      "hub_repo": "google/gemma-4-E4B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 9,
      "disk_mib": 4496,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E4B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.5546
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 45.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6338,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.233,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 303.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.5838,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 61.44,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 3.17,
          "first_at": "2026-10-23T15:22:20.859Z",
          "last_at": "2026-10-23T15:22:20.859Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.5546,
          "usd_per_m_energy": 0.0881,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 42.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7096,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 43.1394,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 25.044,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 153.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6882,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E4B-it-GGUF:Q3_K_M",
      "family": "gemma-4",
      "member": "e4b",
      "quant": "Q3_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E4B-it",
      "hub_repo": "google/gemma-4-E4B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 5,
      "disk_mib": 3380,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E4B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W35"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 60.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2133,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 14.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.3674,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 403.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1884,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 139.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2377,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 56.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.0208,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35"
          ],
          "tokens_per_sec_best": 20.35,
          "effective_ctx_best": 4096,
          "joules_per_token_best": null,
          "first_at": "2026-08-26T10:04:27.273Z",
          "last_at": "2026-08-26T10:04:27.273Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 32.0534,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 18.6169,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 204.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2647,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E4B-it-GGUF:Q4_0",
      "family": "gemma-4",
      "member": "e4b",
      "quant": "Q4_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E4B-it",
      "hub_repo": "google/gemma-4-E4B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 5,
      "disk_mib": 3380,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E4B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.6067
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 60.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2133,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 14.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.3674,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 403.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1884,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 54.67,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 3.007,
          "first_at": "2026-10-23T15:29:06.075Z",
          "last_at": "2026-10-23T15:29:06.075Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.6067,
          "usd_per_m_energy": 0.0835,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 56.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.0208,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.3,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 32.0534,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 18.6169,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 204.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2647,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E4B-it-GGUF:Q4_K_M",
      "family": "gemma-4",
      "member": "e4b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E4B-it",
      "hub_repo": "google/gemma-4-E4B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 4746,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E4B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.6049
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7295,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.4348,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 287.22,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6726,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 56.67,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 2.991,
          "first_at": "2026-10-23T01:25:32.668Z",
          "last_at": "2026-10-23T01:25:32.668Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read"
          ],
          "usd_per_m_out": 0.6049,
          "usd_per_m_energy": 0.0831,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 40.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.8654,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 45.6308,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.82,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 26.5371,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 145.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7834,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gemma-4-E4B-it-GGUF:Q8_0",
      "family": "gemma-4",
      "member": "e4b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "google/gemma-4-E4B-it",
      "hub_repo": "google/gemma-4-E4B-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 1,
      "disk_mib": 7813,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/google/gemma-4-E4B-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.9312
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 26.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.9414,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.0685,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 174.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.7677,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 39.49,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 4.173,
          "first_at": "2026-10-23T06:04:49.084Z",
          "last_at": "2026-10-24T14:52:59.345Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read",
            "vision.spatial"
          ],
          "usd_per_m_out": 0.9312,
          "usd_per_m_energy": 0.1159,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 24.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.8195,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 77.3305,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 45.2225,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 88.6,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.964,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q2_K",
      "family": "gpt-oss",
      "member": "20b",
      "quant": "Q2_K",
      "engine": "ollama",
      "modality": "text",
      "base": "openai/gpt-oss-20b",
      "hub_repo": "openai/gpt-oss-20b",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 10937,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/gpt-oss-20b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": {
        "openrouter_id": "openai/gpt-oss-20b",
        "ids": [
          "openai/gpt-oss-20b",
          "openai/gpt-oss-20b:batch"
        ],
        "usd_per_m_in": 0.03,
        "usd_per_m_out": 0.13,
        "usd_per_m_cached_in": 0.03,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.112,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 9.339,
        "vs_api": 71.84
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 18.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.2531,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": 4.21,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 9.96,
          "first_at": "2026-10-28T10:13:37.125Z",
          "last_at": "2026-11-12T18:04:57.546Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.es",
            "lang.gu",
            "lang.hi",
            "lang.ja",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ne",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.ta",
            "lang.te",
            "lang.ur",
            "lang.zh",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 9.339,
          "usd_per_m_energy": 0.2767,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 124.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.8946,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.13,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9116,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.8895,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 112.4969,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.22,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 65.4695,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 63.29,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.1895,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q3_K_S",
      "family": "gpt-oss",
      "member": "20b",
      "quant": "Q3_K_S",
      "engine": "ollama",
      "modality": "text",
      "base": "openai/gpt-oss-20b",
      "hub_repo": "openai/gpt-oss-20b",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 2,
      "disk_mib": 10932,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/gpt-oss-20b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": {
        "openrouter_id": "openai/gpt-oss-20b",
        "ids": [
          "openai/gpt-oss-20b",
          "openai/gpt-oss-20b:batch"
        ],
        "usd_per_m_in": 0.03,
        "usd_per_m_out": 0.13,
        "usd_per_m_cached_in": 0.03,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.112,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.7183,
        "vs_api": 5.53
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 18.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.2506,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": 4.38,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 13.347,
          "first_at": "2026-10-28T09:08:57.712Z",
          "last_at": "2026-11-12T06:10:29.767Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.es",
            "lang.gu",
            "lang.hi",
            "lang.ja",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ne",
            "lang.nl",
            "lang.or",
            "lang.pa",
            "lang.ru",
            "lang.ta",
            "lang.te",
            "lang.ur",
            "lang.zh",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 8.9756,
          "usd_per_m_energy": 0.3708,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 124.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.893,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 54.73,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 6.178,
          "first_at": "2026-11-03T04:12:22.645Z",
          "last_at": "2026-11-03T04:12:22.645Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 0.7183,
          "usd_per_m_energy": 0.1716,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.8893,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 112.4911,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 64.9339,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 63.32,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.1874,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_0",
      "family": "gpt-oss",
      "member": "20b",
      "quant": "Q4_0",
      "engine": "ollama",
      "modality": "text",
      "base": "openai/gpt-oss-20b",
      "hub_repo": "openai/gpt-oss-20b",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 3,
      "disk_mib": 10968,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/gpt-oss-20b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "openai/gpt-oss-20b",
        "ids": [
          "openai/gpt-oss-20b",
          "openai/gpt-oss-20b:batch"
        ],
        "usd_per_m_in": 0.03,
        "usd_per_m_out": 0.13,
        "usd_per_m_cached_in": 0.03,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.112,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 8.6277,
        "vs_api": 66.37
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 18.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.2681,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W44",
            "2026-W45"
          ],
          "tokens_per_sec_best": 4.56,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 11.745,
          "first_at": "2026-10-28T11:18:09.706Z",
          "last_at": "2026-11-08T19:38:32.437Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.es",
            "lang.hi",
            "lang.ja",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ne",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.ta",
            "lang.te",
            "lang.ur",
            "lang.zh",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 8.6277,
          "usd_per_m_energy": 0.3262,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 124.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.9061,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 43.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9147,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.43,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 6.9107,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.71,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 112.5324,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.22,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 65.4902,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 63.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.2018,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_M",
      "family": "gpt-oss",
      "member": "20b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "openai/gpt-oss-20b",
      "hub_repo": "openai/gpt-oss-20b",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 11086,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/gpt-oss-20b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "openai/gpt-oss-20b",
        "ids": [
          "openai/gpt-oss-20b",
          "openai/gpt-oss-20b:batch"
        ],
        "usd_per_m_in": 0.03,
        "usd_per_m_out": 0.13,
        "usd_per_m_cached_in": 0.03,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.112,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.7528,
        "vs_api": 5.79
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 18.53,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.317,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W44",
            "2026-W45"
          ],
          "tokens_per_sec_best": 4.41,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 11.123,
          "first_at": "2026-10-28T04:29:17.791Z",
          "last_at": "2026-11-04T18:20:14.431Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.as",
            "lang.bn",
            "lang.en",
            "lang.es",
            "lang.gu",
            "lang.hi",
            "lang.ja",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ms",
            "lang.ne",
            "lang.nl",
            "lang.or",
            "lang.pa",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "lang.ur",
            "lang.zh",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 8.943,
          "usd_per_m_energy": 0.309,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 122.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.9488,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 52.39,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 9.638,
          "first_at": "2026-11-02T22:30:34.488Z",
          "last_at": "2026-11-02T22:30:34.488Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 0.7528,
          "usd_per_m_energy": 0.2677,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 24.03,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 3.634,
          "first_at": "2026-11-06T11:50:09.364Z",
          "last_at": "2026-11-06T11:50:09.364Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 5.0166,
          "usd_per_m_energy": 0.1009,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 114.2772,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.21,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 66.1108,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 62.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.2484,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q8_0",
      "family": "gpt-oss",
      "member": "20b",
      "quant": "Q8_0",
      "engine": "ollama",
      "modality": "text",
      "base": "openai/gpt-oss-20b",
      "hub_repo": "openai/gpt-oss-20b",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 1,
      "disk_mib": 11548,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/gpt-oss-20b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "openai/gpt-oss-20b",
        "ids": [
          "openai/gpt-oss-20b",
          "openai/gpt-oss-20b:batch"
        ],
        "usd_per_m_in": 0.03,
        "usd_per_m_out": 0.13,
        "usd_per_m_cached_in": 0.03,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.112,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 9.8551,
        "vs_api": 75.81
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 17.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.5202,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W45"
          ],
          "tokens_per_sec_best": 4.04,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 16.214,
          "first_at": "2026-10-28T12:20:38.040Z",
          "last_at": "2026-11-04T22:34:14.316Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.ha",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.sw",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 9.8551,
          "usd_per_m_energy": 0.4504,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 118.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.1166,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 40.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.9747,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.55,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 7.3067,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.67,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 119.9552,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.16,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 69.2845,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 59.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.4319,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/gpt-oss-20b-GGUF:Q8_K_XL",
      "family": "gpt-oss",
      "member": "20b",
      "quant": "Q8_K_XL",
      "engine": "ollama",
      "modality": "text",
      "base": "openai/gpt-oss-20b",
      "hub_repo": "openai/gpt-oss-20b",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 12584,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/gpt-oss-20b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "openai/gpt-oss-20b",
        "ids": [
          "openai/gpt-oss-20b",
          "openai/gpt-oss-20b:batch"
        ],
        "usd_per_m_in": 0.03,
        "usd_per_m_out": 0.13,
        "usd_per_m_cached_in": 0.03,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.112,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": {
        "class": "linux/amd64/nvidia/32g",
        "usd_per_m_out": 0.9207,
        "vs_api": 7.08
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.32,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.9763,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 3.11,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 26.495,
          "first_at": "2026-11-04T11:42:52.579Z",
          "last_at": "2026-11-04T11:42:52.579Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 13.0732,
          "usd_per_m_energy": 0.736,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 108.32,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.4938,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 44.16,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 5.215,
          "first_at": "2026-11-03T03:58:00.527Z",
          "last_at": "2026-11-03T03:58:00.527Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools.multi_hop",
            "tools_loop_terminated"
          ],
          "usd_per_m_out": 0.9207,
          "usd_per_m_energy": 0.1449,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.0164,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.62,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 130.9889,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 76.6161,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 55.01,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.8444,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/granite-4.1-3b-GGUF:Q4_K_M",
      "family": "granite-4-1",
      "member": "3b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "ibm-granite/granite-4.1-3b",
      "hub_repo": "ibm-granite/granite-4.1-3b",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 2002,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ibm-granite/granite-4.1-3b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 6.2345
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 102.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7078,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 23.66,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.3544,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 680.9,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7022,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 235.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.136,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 18.15,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 28.363,
          "first_at": "2026-11-06T15:18:52.166Z",
          "last_at": "2026-11-06T15:18:52.166Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.kn",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "usd_per_m_out": 6.2345,
          "usd_per_m_energy": 0.7879,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.89,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 18.6636,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 6.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 10.8522,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 345.75,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7459,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/unsloth/medgemma-1.5-4b-it-GGUF:Q4_K_M",
      "family": "medgemma-1-5",
      "member": "4b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "google/medgemma-1.5-4b-it",
      "hub_repo": "google/medgemma-1.5-4b-it",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 2374,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "health-ai-developer-foundations",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/medgemma-1.5-4b-it",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/nvidia/64g",
        "usd_per_m_out": 1.5225
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 86.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8427,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 19.95,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.6215,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 574.2,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8333,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 198.69,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1628,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": 74.52,
          "effective_ctx_best": 65536,
          "joules_per_token_best": 1.714,
          "first_at": "2026-11-07T00:30:11.108Z",
          "last_at": "2026-11-07T00:30:11.108Z",
          "caps": [
            "code.reason",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output",
            "vision",
            "vision.count",
            "vision.multi_image",
            "vision.read"
          ],
          "usd_per_m_out": 1.5225,
          "usd_per_m_energy": 0.0476,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.28,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 22.2269,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.64,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 12.9263,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 291.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.8855,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/victor/functiongemma-agent-gguf:Q4_K_M",
      "family": "functiongemma",
      "member": "270m~unsloth",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "unsloth/functiongemma-270m-it",
      "hub_repo": "unsloth/functiongemma-270m-it",
      "publisher": "victor",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 241,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/unsloth/functiongemma-270m-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": {
        "class": "linux/amd64/none/32g",
        "usd_per_m_out": 0.8341
      },
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 852.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0835,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 36.7,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.257,
          "first_at": "2026-10-25T03:18:21.416Z",
          "last_at": "2026-10-25T03:18:21.416Z",
          "caps": [
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking"
          ],
          "usd_per_m_out": 0.8341,
          "usd_per_m_energy": 0.0349,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5656.24,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0843,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1957.23,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0156,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 793.12,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.1409,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 32.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.2013,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 55.58,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.2805,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2872.18,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0893,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/vultr/Step-3.7-Flash-GGUF:Q4_K_M",
      "family": "step-3-7",
      "member": "flash",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "stepfun-ai/Step-3.7-Flash",
      "hub_repo": "stepfun-ai/Step-3.7-Flash",
      "publisher": "vultr",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 124366,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "stepfun/step-3.7-flash",
        "ids": [
          "stepfun/step-3.7-flash"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 1.15,
        "usd_per_m_cached_in": 0.04,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.65,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 104.3524,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 346.3841,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 10.96,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 52.7132,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 34.7298,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.54,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 138.141,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2869.6922,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.11,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1565.2867,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 5.57,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 64.1758,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "hf.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF:Q3_K_M",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 16.98,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 6.114,
          "first_at": "2026-10-23T17:12:38.055Z",
          "last_at": "2026-10-23T17:12:38.055Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF:Q4_K_M",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 25.81,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 6.184,
          "first_at": "2026-10-23T14:12:37.709Z",
          "last_at": "2026-10-24T17:45:14.364Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.am",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.nl",
            "lang.pa",
            "lang.ru",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF:Q6_K",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 19.96,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 7.624,
          "first_at": "2026-10-23T14:41:23.604Z",
          "last_at": "2026-10-23T14:41:23.604Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "hf.co/yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF:Q8_0",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "yuxinlu1/gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": 18.03,
          "effective_ctx_best": 131072,
          "joules_per_token_best": 9.092,
          "first_at": "2026-10-23T14:55:24.559Z",
          "last_at": "2026-10-23T14:55:24.559Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "huihui-ai/Huihui-GLM-4.7-Flash-abliterated-mlx-4bit",
      "family": "glm-4-7",
      "member": "flash~huihui-ai-abliterated",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-4.7-Flash",
      "hub_repo": "zai-org/GLM-4.7-Flash",
      "publisher": "huihui-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 16071,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.7-Flash",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.7-flash",
        "ids": [
          "z-ai/glm-4.7-flash"
        ],
        "usd_per_m_in": 0.0605,
        "usd_per_m_out": 0.4,
        "usd_per_m_cached_in": null,
        "context_length": 200000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "hunyuanvideo-community/HunyuanVideo",
      "family": "hunyuan-video",
      "member": "t2v-720p",
      "quant": "BF16",
      "engine": "diffusers",
      "modality": "image",
      "base": null,
      "hub_repo": "hunyuanvideo-community/HunyuanVideo",
      "publisher": "hunyuanvideo-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 39962,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/hunyuanvideo-community/HunyuanVideo",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-23T20:01:00.857Z",
          "last_at": "2026-10-23T20:01:00.857Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "ibm-granite/granite-4.2-8b-fp8",
      "family": "granite-4-2",
      "member": "8b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "ibm-granite/granite-4.2-8b",
      "hub_repo": "ibm-granite/granite-4.2-8b",
      "publisher": "ibm-granite",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 9171,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ibm-granite/granite-4.2-8b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "ibm-granite/granite-4.2-8b",
        "ids": [
          "ibm-granite/granite-4.2-8b"
        ],
        "usd_per_m_in": 0.06,
        "usd_per_m_out": 0.15,
        "usd_per_m_cached_in": 0.015,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 54.99,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.1639,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "illitan/Hy-MT2-1.8B-4bit",
      "family": "hy-mt-2",
      "member": "1.8b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "tencent/Hy-MT2-1.8B",
      "hub_repo": "tencent/Hy-MT2-1.8B",
      "publisher": "illitan",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 961,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy-MT2-1.8B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy-mt2-1.8b",
        "ids": [
          "tencent/hy-mt2-1.8b"
        ],
        "usd_per_m_in": 0.044,
        "usd_per_m_out": 0.177,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "kaushikvira/DeepSeek-V4.1-Flash-1017-GSQ-agentic-strata",
      "family": "deepseek-v4-1",
      "member": "flash-1017~kaushikvira-gsq-agentic-strata",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "hub_repo": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "publisher": "kaushikvira",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 193365,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4.1-Flash-1017",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4.1-flash-1017",
        "ids": [
          "deepseek/deepseek-v4.1-flash-1017",
          "deepseek/deepseek-v4.1-flash-1017:batch"
        ],
        "usd_per_m_in": 0.02,
        "usd_per_m_out": 0.108,
        "usd_per_m_cached_in": 0.007,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.336,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 103.0225,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "kernelpool/Kimi-K3-2bit-UVMAX",
      "family": "kimi-k3",
      "member": "base~kernelpool-uvmax",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K3",
      "hub_repo": "moonshotai/Kimi-K3",
      "publisher": "kernelpool",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 778916,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "kimi-k3",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K3",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k3",
        "ids": [
          "moonshotai/kimi-k3",
          "moonshotai/kimi-k3:batch"
        ],
        "usd_per_m_in": 0.192,
        "usd_per_m_out": 16,
        "usd_per_m_cached_in": 0.182,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 11.4,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "kokoro",
      "family": "kokoro",
      "member": "base",
      "quant": null,
      "engine": "speech-pod",
      "modality": "speech",
      "base": null,
      "hub_repo": null,
      "publisher": "kokoro",
      "derivative": false,
      "cohort": 1,
      "rung": 9,
      "disk_mib": 312,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W36",
        "2026-W38",
        "2026-W39",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:01:26.321Z",
          "last_at": "2026-11-12T18:25:57.138Z",
          "caps": [
            "audio",
            "audio.out",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-19T21:12:26.661Z",
          "last_at": "2026-10-20T09:29:43.551Z",
          "caps": [
            "audio",
            "audio.out",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T21:18:23.009Z",
          "last_at": "2026-11-03T21:18:23.009Z",
          "caps": [
            "audio",
            "audio.out"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38",
            "2026-W39",
            "2026-W44",
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-18T14:13:42.394Z",
          "last_at": "2026-11-04T00:33:59.425Z",
          "caps": [
            "audio",
            "audio.out",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W36"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-05T11:04:18.111Z",
          "last_at": "2026-09-05T11:04:18.111Z",
          "caps": [
            "audio"
          ],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "lapadatbogdan/GLM-5.3-Flash-Coder-RCO-152E-GGUF:Q4_K_M",
      "family": "glm-5-3",
      "member": "flash~lapadatbogdan-coder-rco-152e",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "zai-org/GLM-5.3-Flash",
      "hub_repo": "zai-org/GLM-5.3-Flash",
      "publisher": "lapadatbogdan",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 97888,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.3-Flash",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.3-flash",
        "ids": [
          "z-ai/glm-5.3-flash",
          "z-ai/glm-5.3-flash:batch"
        ],
        "usd_per_m_in": 0.0399,
        "usd_per_m_out": 0.0784,
        "usd_per_m_cached_in": 0.03,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.18,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "lightx2v/Qwen-Image-2.1-Lightning",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "lightx2v/Qwen-Image-2.1-Lightning",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T06:30:51.338Z",
          "last_at": "2026-10-21T06:30:51.338Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "llama3.2:1b",
      "family": "llama-3-2",
      "member": "1b",
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": null,
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W34",
        "2026-W35",
        "2026-W36",
        "2026-W37",
        "2026-W38",
        "2026-W39",
        "2026-W40",
        "2026-W41",
        "2026-W42",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/amd64/apple/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38",
            "2026-W39",
            "2026-W40",
            "2026-W41",
            "2026-W42",
            "2026-W43"
          ],
          "tokens_per_sec_best": 3.85,
          "effective_ctx_best": null,
          "joules_per_token_best": 5.578,
          "first_at": "2026-09-16T00:36:50.962Z",
          "last_at": "2026-10-24T00:42:27.381Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "darwin/amd64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W36",
            "2026-W37"
          ],
          "tokens_per_sec_best": 4.07,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-02T10:30:06.382Z",
          "last_at": "2026-09-09T11:28:49.340Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "verdict": "only the network runs it"
        },
        "darwin/arm64/apple/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W41"
          ],
          "tokens_per_sec_best": 20.37,
          "effective_ctx_best": 32768,
          "joules_per_token_best": null,
          "first_at": "2026-10-08T12:18:49.150Z",
          "last_at": "2026-10-08T12:18:49.150Z",
          "caps": [
            "code.write",
            "lang.en",
            "lang.hi",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "verdict": "only the network runs it"
        },
        "darwin/arm64/none/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W36"
          ],
          "tokens_per_sec_best": 25.97,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-06T15:31:24.556Z",
          "last_at": "2026-09-06T15:31:24.556Z",
          "caps": [
            "code.write",
            "lang.en",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "verdict": "only the network runs it"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 3,
          "seasons": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": 11.45,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 3.186,
          "first_at": "2026-10-24T21:02:12.065Z",
          "last_at": "2026-11-10T17:09:26.291Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.nl",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "unknown": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W34",
            "2026-W35"
          ],
          "tokens_per_sec_best": 13.09,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 0.432,
          "first_at": "2026-08-20T07:30:32.000Z",
          "last_at": "2026-08-24T07:11:09.129Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "verdict": "only the network runs it"
        },
        "windows/amd64/amd/8g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38",
            "2026-W44"
          ],
          "tokens_per_sec_best": 7.7,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 17.518,
          "first_at": "2026-09-15T04:52:17.198Z",
          "last_at": "2026-10-28T11:00:35.688Z",
          "caps": [
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.en",
            "lang.hi",
            "lang.nl",
            "output.long",
            "output.schema",
            "structured_output",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.loop_ends",
            "tools_loop_terminated"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "lmstudio-community/MiniMax-M2.5-MLX-4bit",
      "family": "minimax-m2-5",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.5",
      "hub_repo": "MiniMaxAI/MiniMax-M2.5",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 122706,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.5",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.5",
        "ids": [
          "minimax/minimax-m2.5"
        ],
        "usd_per_m_in": 0.27,
        "usd_per_m_out": 0.95,
        "usd_per_m_cached_in": 0.03,
        "context_length": 198000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "lmstudio-community/Qwen3-Coder-30B-A3B-Instruct-MLX-4bit",
      "family": "qwen-3",
      "member": "coder-30b-a3b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3-Coder-30B-A3B-Instruct",
      "hub_repo": "Qwen/Qwen3-Coder-30B-A3B-Instruct",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 16379,
      "active_share": 0.1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-Coder-30B-A3B-Instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3-coder-30b-a3b-instruct",
        "ids": [
          "qwen/qwen3-coder-30b-a3b-instruct"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "lmstudio-community/Qwen3-Coder-Next-MLX-4bit",
      "family": "qwen-3",
      "member": "coder-next",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3-Coder-Next",
      "hub_repo": "Qwen/Qwen3-Coder-Next",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 42766,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3-Coder-Next",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3-coder-next",
        "ids": [
          "qwen/qwen3-coder-next"
        ],
        "usd_per_m_in": 0.12,
        "usd_per_m_out": 0.8,
        "usd_per_m_cached_in": 0.07,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "lmstudio-community/Qwen3.5-9B-MLX-4bit",
      "family": "qwen-3-5",
      "member": "9b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3.5-9B",
      "hub_repo": "Qwen/Qwen3.5-9B",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 5674,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.5-9B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.5-9b",
        "ids": [
          "qwen/qwen3.5-9b"
        ],
        "usd_per_m_in": 0.037,
        "usd_per_m_out": 0.054,
        "usd_per_m_cached_in": 0.033,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "lmstudio-community/Qwen3.6-27B-MLX-4bit",
      "family": "qwen-3-6",
      "member": "27b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3.6-27B",
      "hub_repo": "Qwen/Qwen3.6-27B",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 15310,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.6-27B",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.6-27b",
        "ids": [
          "qwen/qwen3.6-27b"
        ],
        "usd_per_m_in": 0.32,
        "usd_per_m_out": 3.2,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "lmstudio-community/Qwen3.6-35B-A3B-MLX-4bit",
      "family": "qwen-3-6",
      "member": "35b-a3b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3.6-35B-A3B",
      "hub_repo": "Qwen/Qwen3.6-35B-A3B",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 19457,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.6-35B-A3B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.6-35b-a3b",
        "ids": [
          "qwen/qwen3.6-35b-a3b"
        ],
        "usd_per_m_in": 0.1,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.025,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "lmstudio-community/Qwen3.8-27B-MLX-4bit",
      "family": "qwen-3-8",
      "member": "27b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 15310,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "lmstudio-community/gpt-oss-safeguard-20b-MLX-MXFP4",
      "family": "gpt-oss",
      "member": "safeguard-20b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "openai/gpt-oss-safeguard-20b",
      "hub_repo": "openai/gpt-oss-safeguard-20b",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 10607,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/gpt-oss-safeguard-20b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "openai/gpt-oss-safeguard-20b",
        "ids": [
          "openai/gpt-oss-safeguard-20b"
        ],
        "usd_per_m_in": 0.075,
        "usd_per_m_out": 0.3,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "lmstudio-community/granite-4.2-8b-MLX-4bit",
      "family": "granite-4-2",
      "member": "8b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "ibm-granite/granite-4.2-8b",
      "hub_repo": "ibm-granite/granite-4.2-8b",
      "publisher": "lmstudio-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 4716,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ibm-granite/granite-4.2-8b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "ibm-granite/granite-4.2-8b",
        "ids": [
          "ibm-granite/granite-4.2-8b"
        ],
        "usd_per_m_in": 0.06,
        "usd_per_m_out": 0.15,
        "usd_per_m_cached_in": 0.015,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "majentik/Mistral-Medium-3.5-128B-MLX-MXFP4",
      "family": "mistral-medium-3-5",
      "member": "128b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "mistralai/Mistral-Medium-3.5-128B",
      "hub_repo": "mistralai/Mistral-Medium-3.5-128B",
      "publisher": "majentik",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 70710,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Medium-3.5-128B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-medium-3-5",
        "ids": [
          "mistralai/mistral-medium-3-5",
          "mistralai/mistral-medium-3-5:batch"
        ],
        "usd_per_m_in": 1.5,
        "usd_per_m_out": 7.5,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": 3.75,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "majentik/Mistral-Small-4-119B-TurboQuant-MLX-4bit",
      "family": "mistral-small-4",
      "member": "119b-2603~majentik-turboquant",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "mistralai/Mistral-Small-4-119B-2603",
      "hub_repo": "mistralai/Mistral-Small-4-119B-2603",
      "publisher": "majentik",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 64614,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Small-4-119B-2603",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-small-2603",
        "ids": [
          "mistralai/mistral-small-2603"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": 0.015,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "microsoft/Phi-4-mini-instruct",
      "family": "phi-4",
      "member": "mini",
      "quant": "BF16",
      "engine": "vllm",
      "modality": "text",
      "base": "microsoft/Phi-4-mini-instruct",
      "hub_repo": "microsoft/Phi-4-mini-instruct",
      "publisher": "microsoft",
      "derivative": false,
      "cohort": 1,
      "rung": 3,
      "disk_mib": 7316,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/Phi-4-mini-instruct",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 2.583,
          "first_at": "2026-10-22T08:26:56.172Z",
          "last_at": "2026-10-22T08:26:56.172Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.en",
            "output.long",
            "output.schema",
            "research.synthesize",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 68.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7041,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "mistralai/Devstral-2-123B-Instruct-2512",
      "family": "devstral-2",
      "member": "123b-instruct-2512",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "mistralai/Devstral-2-123B-Instruct-2512",
      "hub_repo": "mistralai/Devstral-2-123B-Instruct-2512",
      "publisher": "mistralai",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 244616,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Devstral-2-123B-Instruct-2512",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/devstral-2512",
        "ids": [
          "mistralai/devstral-2512"
        ],
        "usd_per_m_in": 0.4,
        "usd_per_m_out": 2,
        "usd_per_m_cached_in": 0.04,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.06,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 150.7751,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "mistralai/Mistral-Medium-3.5-128B",
      "family": "mistral-medium-3-5",
      "member": "128b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "mistralai/Mistral-Medium-3.5-128B",
      "hub_repo": "mistralai/Mistral-Medium-3.5-128B",
      "publisher": "mistralai",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 254833,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Medium-3.5-128B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-medium-3-5",
        "ids": [
          "mistralai/mistral-medium-3-5",
          "mistralai/mistral-medium-3-5:batch"
        ],
        "usd_per_m_in": 1.5,
        "usd_per_m_out": 7.5,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": 3.75,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.98,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 161.0664,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "mistralai/Mistral-Small-4-119B-2603",
      "family": "mistral-small-4",
      "member": "119b-2603",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "mistralai/Mistral-Small-4-119B-2603",
      "hub_repo": "mistralai/Mistral-Small-4-119B-2603",
      "publisher": "mistralai",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 230646,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Small-4-119B-2603",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-small-2603",
        "ids": [
          "mistralai/mistral-small-2603"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": 0.015,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.19,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 136.6338,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "mlx-community/DeepSeek-V3.1-4bit",
      "family": "deepseek-v3-1",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V3.1",
      "hub_repo": "deepseek-ai/DeepSeek-V3.1",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 360114,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.1",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-chat-v3.1",
        "ids": [
          "deepseek/deepseek-chat-v3.1"
        ],
        "usd_per_m_in": 0.25,
        "usd_per_m_out": 0.95,
        "usd_per_m_cached_in": 0.13,
        "context_length": 163840,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/DeepSeek-V3.1-Terminus-4bit",
      "family": "deepseek-v3-1",
      "member": "terminus",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V3.1-Terminus",
      "hub_repo": "deepseek-ai/DeepSeek-V3.1-Terminus",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 360114,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.1-Terminus",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v3.1-terminus",
        "ids": [
          "deepseek/deepseek-v3.1-terminus"
        ],
        "usd_per_m_in": 0.27,
        "usd_per_m_out": 1,
        "usd_per_m_cached_in": null,
        "context_length": 163840,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/DeepSeek-V3.1-mlx-DQ5_K_M",
      "family": "deepseek-v3-1",
      "member": "base~mlx-community-dq5-k-m",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V3.1",
      "hub_repo": "deepseek-ai/DeepSeek-V3.1",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 451008,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.1",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-chat-v3.1",
        "ids": [
          "deepseek/deepseek-chat-v3.1"
        ],
        "usd_per_m_in": 0.25,
        "usd_per_m_out": 0.95,
        "usd_per_m_cached_in": 0.13,
        "context_length": 163840,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/DeepSeek-V3.2-4bit",
      "family": "deepseek-v3-2",
      "member": "base~mlx-community",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V3.2",
      "hub_repo": "deepseek-ai/DeepSeek-V3.2",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 360571,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.2",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v3.2",
        "ids": [
          "deepseek/deepseek-v3.2"
        ],
        "usd_per_m_in": 0.26,
        "usd_per_m_out": 0.38,
        "usd_per_m_cached_in": 0.13,
        "context_length": 163840,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/DeepSeek-V4-Flash-4bit",
      "family": "deepseek-v4",
      "member": "flash~mlx-community",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": null,
      "hub_repo": "mlx-community/DeepSeek-V4-Flash-4bit",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 144465,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4-Flash",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4-flash",
        "ids": [
          "deepseek/deepseek-v4-flash"
        ],
        "usd_per_m_in": 0.09,
        "usd_per_m_out": 0.18,
        "usd_per_m_cached_in": 0.018,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Devstral-2-123B-Instruct-2512-4bit",
      "family": "devstral-2",
      "member": "123b-instruct-2512",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "mistralai/Devstral-2-123B-Instruct-2512",
      "hub_repo": "mistralai/Devstral-2-123B-Instruct-2512",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 67072,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Devstral-2-123B-Instruct-2512",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/devstral-2512",
        "ids": [
          "mistralai/devstral-2512"
        ],
        "usd_per_m_in": 0.4,
        "usd_per_m_out": 2,
        "usd_per_m_cached_in": 0.04,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/GLM-4.5-4bit",
      "family": "glm-4-5",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-4.5",
      "hub_repo": "zai-org/GLM-4.5",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 189359,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.5",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.5",
        "ids": [
          "z-ai/glm-4.5"
        ],
        "usd_per_m_in": 0.6,
        "usd_per_m_out": 2.2,
        "usd_per_m_cached_in": 0.11,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/GLM-4.5-Air-4bit",
      "family": "glm-4-5",
      "member": "air",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-4.5-Air",
      "hub_repo": "zai-org/GLM-4.5-Air",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 57353,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.5-Air",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.5-air",
        "ids": [
          "z-ai/glm-4.5-air"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 1.1,
        "usd_per_m_cached_in": 0.03,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/GLM-4.5V-4bit",
      "family": "glm-4-5",
      "member": "v~mlx-community",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": null,
      "hub_repo": "mlx-community/GLM-4.5V-4bit",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 58991,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mlx-community/GLM-4.5V-4bit",
        "fetched_at": "2026-11-14T09:30:52.417Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.5v",
        "ids": [
          "z-ai/glm-4.5v"
        ],
        "usd_per_m_in": 0.6,
        "usd_per_m_out": 1.8,
        "usd_per_m_cached_in": 0.11,
        "context_length": 65536,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/GLM-4.6-4bit",
      "family": "glm-4-6",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-4.6",
      "hub_repo": "zai-org/GLM-4.6",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 189359,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.6",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.6",
        "ids": [
          "z-ai/glm-4.6"
        ],
        "usd_per_m_in": 0.5,
        "usd_per_m_out": 2,
        "usd_per_m_cached_in": 0.1,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/GLM-4.6V-4bit",
      "family": "glm-4-6",
      "member": "v~mlx-community",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": null,
      "hub_repo": "mlx-community/GLM-4.6V-4bit",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 58991,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mlx-community/GLM-4.6V-4bit",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.6v",
        "ids": [
          "z-ai/glm-4.6v"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.055,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/GLM-4.6V-mxfp4",
      "family": "glm-4-6",
      "member": "v",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-4.6V",
      "hub_repo": "zai-org/GLM-4.6V",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 55808,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.6V",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.6v",
        "ids": [
          "z-ai/glm-4.6v"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.055,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/GLM-4.7-4bit",
      "family": "glm-4-7",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-4.7",
      "hub_repo": "zai-org/GLM-4.7",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 189359,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.7",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.7",
        "ids": [
          "z-ai/glm-4.7"
        ],
        "usd_per_m_in": 0.4,
        "usd_per_m_out": 1.75,
        "usd_per_m_cached_in": 0.08,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/GLM-4.7-Flash-4bit",
      "family": "glm-4-7",
      "member": "flash",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-4.7-Flash",
      "hub_repo": "zai-org/GLM-4.7-Flash",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 16071,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.7-Flash",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.7-flash",
        "ids": [
          "z-ai/glm-4.7-flash"
        ],
        "usd_per_m_in": 0.0605,
        "usd_per_m_out": 0.4,
        "usd_per_m_cached_in": null,
        "context_length": 200000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/GLM-5-4bit",
      "family": "glm-5",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5",
      "hub_repo": "zai-org/GLM-5",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 399228,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5",
        "ids": [
          "z-ai/glm-5"
        ],
        "usd_per_m_in": 1,
        "usd_per_m_out": 3.2,
        "usd_per_m_cached_in": 0.2,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/GLM-5.1-DQ4plus-q8",
      "family": "glm-5-1",
      "member": "base~mlx-community-dq4plus",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.1",
      "hub_repo": "zai-org/GLM-5.1",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 443624,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.1",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.1",
        "ids": [
          "z-ai/glm-5.1"
        ],
        "usd_per_m_in": 1.4,
        "usd_per_m_out": 4.4,
        "usd_per_m_cached_in": 0.26,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/GLM-5.2-DQ4plus-q8",
      "family": "glm-5-2",
      "member": "base~mlx-community-dq4plus",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.2",
      "hub_repo": "zai-org/GLM-5.2",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 443083,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.2",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.2",
        "ids": [
          "z-ai/glm-5.2"
        ],
        "usd_per_m_in": 0.007,
        "usd_per_m_out": 6.8,
        "usd_per_m_cached_in": 0.006,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/GLM-5.2-mxfp4",
      "family": "glm-5-2",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.2-FP8",
      "hub_repo": "zai-org/GLM-5.2-FP8",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 376791,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.2-FP8",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.2",
        "ids": [
          "z-ai/glm-5.2"
        ],
        "usd_per_m_in": 0.007,
        "usd_per_m_out": 6.8,
        "usd_per_m_cached_in": 0.006,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Huihui-GLM-4.5V-abliterated-mxfp4",
      "family": "glm-4-5",
      "member": "v~mlx-community-huihui-abliterated",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-4.5V",
      "hub_repo": "zai-org/GLM-4.5V",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 55808,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.5V",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.5v",
        "ids": [
          "z-ai/glm-4.5v"
        ],
        "usd_per_m_in": 0.6,
        "usd_per_m_out": 1.8,
        "usd_per_m_cached_in": 0.11,
        "context_length": 65536,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Hy-MT2-1.8B-4bit",
      "family": "hy-mt-2",
      "member": "1.8b~mlx-community",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": null,
      "hub_repo": "mlx-community/Hy-MT2-1.8B-4bit",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 961,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mlx-community/Hy-MT2-1.8B-4bit",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy-mt2-1.8b",
        "ids": [
          "tencent/hy-mt2-1.8b"
        ],
        "usd_per_m_in": 0.044,
        "usd_per_m_out": 0.177,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Hy-MT2-7B-4bit",
      "family": "hy-mt-2",
      "member": "7b~mlx-community",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": null,
      "hub_repo": "mlx-community/Hy-MT2-7B-4bit",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 4026,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mlx-community/Hy-MT2-7B-4bit",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy-mt2-7b",
        "ids": [
          "tencent/hy-mt2-7b"
        ],
        "usd_per_m_in": 0.074,
        "usd_per_m_out": 0.295,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Hy3-OptiQ-2bit",
      "family": "hy-3",
      "member": "base~mlx-community-optiq",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "tencent/Hy3",
      "hub_repo": "tencent/Hy3",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 84043,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy3",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy3",
        "ids": [
          "tencent/hy3"
        ],
        "usd_per_m_in": 0.0825,
        "usd_per_m_out": 0.33,
        "usd_per_m_cached_in": 0.020625,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Hy3-preview-4bit",
      "family": "hy-3",
      "member": "preview",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "tencent/Hy3-preview",
      "hub_repo": "tencent/Hy3-preview",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 158299,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy3-preview",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy3-preview",
        "ids": [
          "tencent/hy3-preview"
        ],
        "usd_per_m_in": 0.18,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": 0.06,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Kimi-K2-Instruct-0905-mlx-DQ3_K_M",
      "family": "kimi-k2",
      "member": "instruct-0905~mlx-community-dq3-k-m",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2-Instruct-0905",
      "hub_repo": "moonshotai/Kimi-K2-Instruct-0905",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 446707,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2-Instruct-0905",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2-0905",
        "ids": [
          "moonshotai/kimi-k2-0905"
        ],
        "usd_per_m_in": 0.6,
        "usd_per_m_out": 2.5,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Kimi-K2-Instruct-4bit",
      "family": "kimi-k2",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2-Instruct",
      "hub_repo": "moonshotai/Kimi-K2-Instruct",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 550836,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2-Instruct",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2",
        "ids": [
          "moonshotai/kimi-k2"
        ],
        "usd_per_m_in": 0.57,
        "usd_per_m_out": 2.3,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Kimi-K2-Thinking",
      "family": "kimi-k2",
      "member": "thinking",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2-Thinking",
      "hub_repo": "moonshotai/Kimi-K2-Thinking",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 627158,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2-Thinking",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2-thinking",
        "ids": [
          "moonshotai/kimi-k2-thinking"
        ],
        "usd_per_m_in": 0.6,
        "usd_per_m_out": 2.5,
        "usd_per_m_cached_in": 0.15,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Kimi-K2.5",
      "family": "kimi-k2-5",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.5",
      "hub_repo": "moonshotai/Kimi-K2.5",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 627158,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.5",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.5",
        "ids": [
          "moonshotai/kimi-k2.5"
        ],
        "usd_per_m_in": 0.5,
        "usd_per_m_out": 2.5,
        "usd_per_m_cached_in": 0.15,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Kimi-K2.5-mlx-DQ3_K_M-q8",
      "family": "kimi-k2-5",
      "member": "base~mlx-community-dq3-k-m",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.5",
      "hub_repo": "moonshotai/Kimi-K2.5",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 448826,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.5",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.5",
        "ids": [
          "moonshotai/kimi-k2.5"
        ],
        "usd_per_m_in": 0.5,
        "usd_per_m_out": 2.5,
        "usd_per_m_cached_in": 0.15,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Kimi-K2.6-mlx-DQ3_K_M-q8",
      "family": "kimi-k2-6",
      "member": "base~mlx-community-dq3-k-m",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.6",
      "hub_repo": "moonshotai/Kimi-K2.6",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 448826,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.6",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.6",
        "ids": [
          "moonshotai/kimi-k2.6"
        ],
        "usd_per_m_in": 0.65,
        "usd_per_m_out": 3.41,
        "usd_per_m_cached_in": 0.15,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Kimi-K2.7-Code-4bit",
      "family": "kimi-k2-7",
      "member": "code~mlx-community",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": null,
      "hub_repo": "mlx-community/Kimi-K2.7-Code-4bit",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 611316,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.7-Code",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.7-code",
        "ids": [
          "moonshotai/kimi-k2.7-code"
        ],
        "usd_per_m_in": 0.7125,
        "usd_per_m_out": 3.5,
        "usd_per_m_cached_in": 0.16,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Kimi-K2.7-Code-mlx-DQ3_K_M-q8",
      "family": "kimi-k2-7",
      "member": "code~mlx-community-dq3-k-m",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.7-Code",
      "hub_repo": "moonshotai/Kimi-K2.7-Code",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 448826,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.7-Code",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.7-code",
        "ids": [
          "moonshotai/kimi-k2.7-code"
        ],
        "usd_per_m_in": 0.7125,
        "usd_per_m_out": 3.5,
        "usd_per_m_cached_in": 0.16,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Llama-3.2-3B-Instruct-4bit",
      "family": "llama-3-2",
      "member": "3b~mlx-community-llama-3-2",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "mlx-community/Llama-3.2-3B-Instruct",
      "hub_repo": "mlx-community/Llama-3.2-3B-Instruct",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 1723,
      "active_share": 1,
      "licence": {
        "id": "llama3.2",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mlx-community/Llama-3.2-3B-Instruct",
        "fetched_at": "2026-11-10T14:31:12.618Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta-llama/llama-3.2-3b-instruct",
        "ids": [
          "meta-llama/llama-3.2-3b-instruct"
        ],
        "usd_per_m_in": 0.0509,
        "usd_per_m_out": 0.335,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/MiMo-V2.6-Flash-RL-mxfp4-q8",
      "family": "mimo-v2-6",
      "member": "flash-rl",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "hub_repo": "XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 159515,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.6-Flash-RL",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.6-flash",
        "ids": [
          "xiaomi/mimo-v2.6-flash"
        ],
        "usd_per_m_in": 0.14,
        "usd_per_m_out": 0.28,
        "usd_per_m_cached_in": 0.0028,
        "context_length": 1050000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/MiMo-V2.6-Pro-RL-mxfp4-q8",
      "family": "mimo-v2-6",
      "member": "pro-rl",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "XiaomiMiMo/MiMo-V2.6-Pro-RL",
      "hub_repo": "XiaomiMiMo/MiMo-V2.6-Pro-RL",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 528219,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.6-Pro-RL",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.6-pro",
        "ids": [
          "xiaomi/mimo-v2.6-pro"
        ],
        "usd_per_m_in": 0.435,
        "usd_per_m_out": 0.87,
        "usd_per_m_cached_in": 0.0036,
        "context_length": 1050000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/MiniMax-M2.7-4bit",
      "family": "minimax-m2-7",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.7",
      "hub_repo": "MiniMaxAI/MiniMax-M2.7",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 122706,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "other",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.7",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.7",
        "ids": [
          "minimax/minimax-m2.7"
        ],
        "usd_per_m_in": 0.21,
        "usd_per_m_out": 0.84,
        "usd_per_m_cached_in": 0.042,
        "context_length": 196608,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/MiniMax-M3-4bit",
      "family": "minimax-m3",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M3",
      "hub_repo": "MiniMaxAI/MiniMax-M3",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 230294,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "minimax-community",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M3",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m3",
        "ids": [
          "minimax/minimax-m3",
          "minimax/minimax-m3:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.06,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.88,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/MiniMax-M3.1-Flash-oQ4e-mtp",
      "family": "minimax-m3-1",
      "member": "flash~mlx-community-oq4e-mtp",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M3.1-Flash",
      "hub_repo": "MiniMaxAI/MiniMax-M3.1-Flash",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 237247,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "minimax-community",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M3.1-Flash",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m3.1-flash",
        "ids": [
          "minimax/minimax-m3.1-flash"
        ],
        "usd_per_m_in": 0.075,
        "usd_per_m_out": 0.3,
        "usd_per_m_cached_in": 0.0075,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Mistral-Medium-3.5-128B-4bit",
      "family": "mistral-medium-3-5",
      "member": "128b~mlx-community",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": null,
      "hub_repo": "mlx-community/Mistral-Medium-3.5-128B-4bit",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 74388,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Medium-3.5-128B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-medium-3-5",
        "ids": [
          "mistralai/mistral-medium-3-5",
          "mistralai/mistral-medium-3-5:batch"
        ],
        "usd_per_m_in": 1.5,
        "usd_per_m_out": 7.5,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": 3.75,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Mistral-Small-4-119B-2603-4bit",
      "family": "mistral-small-4",
      "member": "119b-2603~mlx-community",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": null,
      "hub_repo": "mlx-community/Mistral-Small-4-119B-2603-4bit",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 64614,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mlx-community/Mistral-Small-4-119B-2603-4bit",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-small-2603",
        "ids": [
          "mistralai/mistral-small-2603"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": 0.015,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Mistral-Small-4-119B-2603-OptiQ-2bit",
      "family": "mistral-small-4",
      "member": "119b-2603~mlx-community-optiq",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "mistralai/Mistral-Small-4-119B-2603",
      "hub_repo": "mistralai/Mistral-Small-4-119B-2603",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 43673,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Small-4-119B-2603",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-small-2603",
        "ids": [
          "mistralai/mistral-small-2603"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": 0.015,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Muse-Glimmer-30B-4bit",
      "family": "muse-glimmer",
      "member": "30b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "meta-models/Muse-Glimmer-30B",
      "hub_repo": "meta-models/Muse-Glimmer-30B",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 18515,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/meta-models/Muse-Glimmer-30B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta/muse-glimmer-30b",
        "ids": [
          "meta/muse-glimmer-30b",
          "meta/muse-glimmer-30b:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.04,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.96,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Muse-Glimmer-30B-OptiQ-4bit",
      "family": "muse-glimmer",
      "member": "30b~mlx-community-optiq",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "meta-models/Muse-Glimmer-30B",
      "hub_repo": "meta-models/Muse-Glimmer-30B",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 22743,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/meta-models/Muse-Glimmer-30B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta/muse-glimmer-30b",
        "ids": [
          "meta/muse-glimmer-30b",
          "meta/muse-glimmer-30b:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.04,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.96,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Muse-Spark-Open-816B-A42B-4bit",
      "family": "muse-spark",
      "member": "816b-a42b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "meta-models/Muse-Spark-Open-816B-A42B",
      "hub_repo": "meta-models/Muse-Spark-Open-816B-A42B",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 508099,
      "active_share": 0.0515,
      "licence": {
        "id": "other",
        "name": "muse-model-license",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-models/Muse-Spark-Open-816B-A42B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta/muse-spark-open",
        "ids": [
          "meta/muse-spark-open"
        ],
        "usd_per_m_in": 1.25,
        "usd_per_m_out": 4.25,
        "usd_per_m_cached_in": 0.15,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Muse-Spark-Open-816B-A42B-oQ4e",
      "family": "muse-spark",
      "member": "816b-a42b~mlx-community-open-oq4e",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "meta-models/Muse-Spark-Open-816B-A42B",
      "hub_repo": "meta-models/Muse-Spark-Open-816B-A42B",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 455481,
      "active_share": 0.0515,
      "licence": {
        "id": "other",
        "name": "muse-model-license",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-models/Muse-Spark-Open-816B-A42B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta/muse-spark-open",
        "ids": [
          "meta/muse-spark-open"
        ],
        "usd_per_m_in": 1.25,
        "usd_per_m_out": 4.25,
        "usd_per_m_cached_in": 0.15,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/North-Mini-Code-1.0-4bit",
      "family": "north-1",
      "member": "mini-code-1-0",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "CohereLabs/North-Mini-Code-1.0",
      "hub_repo": "CohereLabs/North-Mini-Code-1.0",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 17638,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/CohereLabs/North-Mini-Code-1.0",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/Step-3.7-Flash-4bit",
      "family": "step-3-7",
      "member": "flash",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "stepfun-ai/Step-3.7-Flash",
      "hub_repo": "stepfun-ai/Step-3.7-Flash",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 105679,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "stepfun/step-3.7-flash",
        "ids": [
          "stepfun/step-3.7-flash"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 1.15,
        "usd_per_m_cached_in": 0.04,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/gpt-oss-120b-MXFP4-Q8",
      "family": "gpt-oss",
      "member": "120b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "openai/gpt-oss-120b",
      "hub_repo": "openai/gpt-oss-120b",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 60450,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/gpt-oss-120b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "openai/gpt-oss-120b",
        "ids": [
          "openai/gpt-oss-120b",
          "openai/gpt-oss-120b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": 0.075,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.136,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/gpt-oss-20b-MXFP4-Q8",
      "family": "gpt-oss",
      "member": "20b",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "openai/gpt-oss-20b",
      "hub_repo": "openai/gpt-oss-20b",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 11516,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/gpt-oss-20b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "openai/gpt-oss-20b",
        "ids": [
          "openai/gpt-oss-20b",
          "openai/gpt-oss-20b:batch"
        ],
        "usd_per_m_in": 0.03,
        "usd_per_m_out": 0.13,
        "usd_per_m_cached_in": 0.03,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.112,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/gpt-oss-20b-OptiQ-4bit",
      "family": "gpt-oss",
      "member": "20b~mlx-community-optiq",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "openai/gpt-oss-20b",
      "hub_repo": "openai/gpt-oss-20b",
      "publisher": "mlx-community",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 11085,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/gpt-oss-20b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "openai/gpt-oss-20b",
        "ids": [
          "openai/gpt-oss-20b",
          "openai/gpt-oss-20b:batch"
        ],
        "usd_per_m_in": 0.03,
        "usd_per_m_out": 0.13,
        "usd_per_m_cached_in": 0.03,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.112,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/mistralai_Ministral-3-14B-Instruct-2512-MLX-MXFP4",
      "family": "ministral-3",
      "member": "14b-instruct-2512",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "mistralai/Ministral-3-14B-Instruct-2512",
      "hub_repo": "mistralai/Ministral-3-14B-Instruct-2512",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 6843,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Ministral-3-14B-Instruct-2512",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/ministral-14b-2512",
        "ids": [
          "mistralai/ministral-14b-2512"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 0.2,
        "usd_per_m_cached_in": 0.02,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mlx-community/phi-4-4bit",
      "family": "phi-4",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "microsoft/phi-4",
      "hub_repo": "microsoft/phi-4",
      "publisher": "mlx-community",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 7864,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/microsoft/phi-4",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "microsoft/phi-4",
        "ids": [
          "microsoft/phi-4"
        ],
        "usd_per_m_in": 0.07,
        "usd_per_m_out": 0.14,
        "usd_per_m_cached_in": null,
        "context_length": 16384,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "moonshotai/Kimi-K2-Instruct",
      "family": "kimi-k2",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "moonshotai/Kimi-K2-Instruct",
      "hub_repo": "moonshotai/Kimi-K2-Instruct",
      "publisher": "moonshotai",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 981513,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2-Instruct",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2",
        "ids": [
          "moonshotai/kimi-k2"
        ],
        "usd_per_m_in": 0.57,
        "usd_per_m_out": 2.3,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1784.8727,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "moonshotai/Kimi-K2-Instruct-0905",
      "family": "kimi-k2",
      "member": "instruct-0905",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "moonshotai/Kimi-K2-Instruct-0905",
      "hub_repo": "moonshotai/Kimi-K2-Instruct-0905",
      "publisher": "moonshotai",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 981513,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2-Instruct-0905",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2-0905",
        "ids": [
          "moonshotai/kimi-k2-0905"
        ],
        "usd_per_m_in": 0.6,
        "usd_per_m_out": 2.5,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1784.8727,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "mrfakename/DeepSeek-V4.1-Flash-1017-GGUF:Q4_K_M",
      "family": "deepseek-v4-1",
      "member": "flash-1017",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "hub_repo": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "publisher": "mrfakename",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 424134,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4.1-Flash-1017",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4.1-flash-1017",
        "ids": [
          "deepseek/deepseek-v4.1-flash-1017",
          "deepseek/deepseek-v4.1-flash-1017:batch"
        ],
        "usd_per_m_in": 0.02,
        "usd_per_m_out": 0.108,
        "usd_per_m_cached_in": 0.007,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.336,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "mrfakename/DeepSeek-V4.1-Flash-1017-abliterated",
      "family": "deepseek-v4-1",
      "member": "flash-1017~mrfakename-abliterated",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "hub_repo": "deepseek-ai/DeepSeek-V4.1-Flash-1017",
      "publisher": "mrfakename",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 486656,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4.1-Flash-1017",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4.1-flash-1017",
        "ids": [
          "deepseek/deepseek-v4.1-flash-1017",
          "deepseek/deepseek-v4.1-flash-1017:batch"
        ],
        "usd_per_m_in": 0.02,
        "usd_per_m_out": 0.108,
        "usd_per_m_cached_in": 0.007,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.336,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.04,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 488.0475,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "nbkcnil/sd-turbo-memorized-commons10",
      "family": "sd-turbo",
      "member": "memorized-commons10",
      "quant": null,
      "engine": "diffusers",
      "modality": "image",
      "base": null,
      "hub_repo": "nbkcnil/sd-turbo-memorized-commons10",
      "publisher": "nbkcnil",
      "derivative": false,
      "cohort": 3,
      "rung": 9,
      "disk_mib": 3269,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/nbkcnil/sd-turbo-memorized-commons10",
        "fetched_at": "2026-11-14T07:14:20.735Z"
      },
      "seasons": [
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-14T22:48:08.543Z",
          "last_at": "2026-11-14T22:48:08.543Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "nm-testing/GLM-4.5-Air-0.72B-MTP",
      "family": "glm-4-5",
      "member": "0.72b~nm-testing-air-mtp",
      "quant": "BF16",
      "engine": "vllm",
      "modality": "text",
      "base": null,
      "hub_repo": "nm-testing/GLM-4.5-Air-0.72B-MTP",
      "publisher": "nm-testing",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 1370,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/nm-testing/GLM-4.5-Air-0.72B-MTP",
        "fetched_at": "2026-11-12T12:44:40.777Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": 0.579,
          "first_at": "2026-11-06T07:33:08.576Z",
          "last_at": "2026-11-06T07:33:08.576Z",
          "caps": [],
          "est_tokens_per_sec": 368.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.306,
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        }
      }
    },
    {
      "ref": "nvidia/DeepSeek-V4-Flash-NVFP4",
      "family": "deepseek-v4",
      "member": "flash",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4-Flash",
      "hub_repo": "deepseek-ai/DeepSeek-V4-Flash",
      "publisher": "nvidia",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 160486,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4-Flash",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4-flash",
        "ids": [
          "deepseek/deepseek-v4-flash"
        ],
        "usd_per_m_in": 0.09,
        "usd_per_m_out": 0.18,
        "usd_per_m_cached_in": 0.018,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.14,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 77.112,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "openai/whisper-large-v3",
      "family": "whisper",
      "member": "large-v3",
      "quant": "FP16",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "openai/whisper-large-v3",
      "hub_repo": "openai/whisper-large-v3",
      "publisher": "openai",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 2945,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/whisper-large-v3",
        "fetched_at": "2026-11-10T15:57:15.731Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:59:49.480Z",
          "last_at": "2026-11-03T22:59:49.480Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.ne",
            "lang.pa",
            "lang.ta",
            "lang.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "openai/whisper-large-v3-turbo",
      "family": "whisper",
      "member": "large-v3-turbo",
      "quant": "FP16",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "openai/whisper-large-v3-turbo",
      "hub_repo": "openai/whisper-large-v3-turbo",
      "publisher": "openai",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 1547,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/whisper-large-v3-turbo",
        "fetched_at": "2026-11-10T15:57:15.731Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-05T05:04:19.701Z",
          "last_at": "2026-11-05T05:06:57.448Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.ne",
            "lang.ta",
            "lang.te",
            "lang.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W43",
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T20:01:26.298Z",
          "last_at": "2026-11-06T18:58:44.856Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.ne",
            "lang.ta",
            "lang.te",
            "lang.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T23:40:50.590Z",
          "last_at": "2026-10-21T23:40:50.590Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T21:17:22.315Z",
          "last_at": "2026-11-06T07:55:25.054Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.ne",
            "lang.ta",
            "lang.te",
            "lang.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "openai/whisper-medium",
      "family": "whisper",
      "member": "medium",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "openai/whisper-medium",
      "hub_repo": "openai/whisper-medium",
      "publisher": "openai",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 2913,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/whisper-medium",
        "fetched_at": "2026-11-10T15:57:15.731Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:38:18.890Z",
          "last_at": "2026-11-03T22:38:18.890Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.en",
            "lang.hi",
            "lang.mr",
            "lang.ta"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-28T08:25:11.444Z",
          "last_at": "2026-10-28T08:25:11.444Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "openai/whisper-small",
      "family": "whisper",
      "member": "small",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "openai/whisper-small",
      "hub_repo": "openai/whisper-small",
      "publisher": "openai",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 922,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/whisper-small",
        "fetched_at": "2026-11-10T15:57:15.731Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W44",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43",
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T22:19:20.313Z",
          "last_at": "2026-10-27T11:10:31.689Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.en",
            "lang.hi",
            "lang.mr",
            "lang.ne",
            "lang.ta",
            "lang.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:35:28.607Z",
          "last_at": "2026-11-03T22:35:28.607Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.en",
            "lang.hi",
            "lang.mr",
            "lang.ne",
            "lang.ta",
            "lang.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-31T01:22:14.407Z",
          "last_at": "2026-10-31T01:22:14.407Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.en",
            "lang.hi"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "optimum-intel-internal-testing/tiny-random-qwen-image",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "optimum-intel-internal-testing/tiny-random-qwen-image",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W38"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-16T16:15:27.204Z",
          "last_at": "2026-09-16T16:15:27.204Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-16T14:50:04.425Z",
          "last_at": "2026-09-16T14:50:04.425Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "orcarouter/GLM-5.3-Flash-MLX",
      "family": "glm-5-3",
      "member": "flash",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.3-Flash",
      "hub_repo": "zai-org/GLM-5.3-Flash",
      "publisher": "orcarouter",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 1082666,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.3-Flash",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.3-flash",
        "ids": [
          "z-ai/glm-5.3-flash",
          "z-ai/glm-5.3-flash:batch"
        ],
        "usd_per_m_in": 0.0399,
        "usd_per_m_out": 0.0784,
        "usd_per_m_cached_in": 0.03,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.18,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "orcarouter/GLM-5.3-MLX",
      "family": "glm-5-3",
      "member": "base",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.3",
      "hub_repo": "zai-org/GLM-5.3",
      "publisher": "orcarouter",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 2173342,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "glm-5.3",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.3",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.3",
        "ids": [
          "z-ai/glm-5.3",
          "z-ai/glm-5.3:batch"
        ],
        "usd_per_m_in": 0.084,
        "usd_per_m_out": 3.6,
        "usd_per_m_cached_in": 0.079,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "orcarouter/Qwen3.8-27B-Uncensored-MLX",
      "family": "qwen-3-8",
      "member": "27b~orcarouter-uncensored",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "orcarouter",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 90187,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "philipjohnbasile/GLM-5.2-Demolition-q4a4-soul-MLX",
      "family": "glm-5-2",
      "member": "base~philipjohnbasile-demolition-soul",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.2",
      "hub_repo": "zai-org/GLM-5.2",
      "publisher": "philipjohnbasile",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 178821,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.2",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.2",
        "ids": [
          "z-ai/glm-5.2"
        ],
        "usd_per_m_in": 0.007,
        "usd_per_m_out": 6.8,
        "usd_per_m_cached_in": 0.006,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "philipjohnbasile/hy3-family-mini-qwen35b-mtp-v1",
      "family": "hy-3",
      "member": "35b-a3b~qwen-qwen3-6",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3.6-35B-A3B",
      "hub_repo": "Qwen/Qwen3.6-35B-A3B",
      "publisher": "philipjohnbasile",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 19058,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.6-35B-A3B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.6-35b-a3b",
        "ids": [
          "qwen/qwen3.6-35b-a3b"
        ],
        "usd_per_m_in": 0.1,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.025,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "pipenetwork/GLM-5.2-REAP50-MLX-4bit",
      "family": "glm-5-2",
      "member": "base~pipenetwork-reap50",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.2",
      "hub_repo": "zai-org/GLM-5.2",
      "publisher": "pipenetwork",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 204429,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.2",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.2",
        "ids": [
          "z-ai/glm-5.2"
        ],
        "usd_per_m_in": 0.007,
        "usd_per_m_out": 6.8,
        "usd_per_m_cached_in": 0.006,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "pipenetwork/GLM-5.3-Flash-MLX-mixed-4_8bit",
      "family": "glm-5-3",
      "member": "flash~pipenetwork-mixed-4",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.3-Flash",
      "hub_repo": "zai-org/GLM-5.3-Flash",
      "publisher": "pipenetwork",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 173496,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.3-Flash",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.3-flash",
        "ids": [
          "z-ai/glm-5.3-flash",
          "z-ai/glm-5.3-flash:batch"
        ],
        "usd_per_m_in": 0.0399,
        "usd_per_m_out": 0.0784,
        "usd_per_m_cached_in": 0.03,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.18,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "pipenetwork/Kimi-K2.7-Code-MLX-4bit-hiprec",
      "family": "kimi-k2-7",
      "member": "code~pipenetwork-hiprec",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.7-Code",
      "hub_repo": "moonshotai/Kimi-K2.7-Code",
      "publisher": "pipenetwork",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 614071,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.7-Code",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.7-code",
        "ids": [
          "moonshotai/kimi-k2.7-code"
        ],
        "usd_per_m_in": 0.7125,
        "usd_per_m_out": 3.5,
        "usd_per_m_cached_in": 0.16,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "prem-research/DeepSeek-V4-Flash-0731-abliterated",
      "family": "deepseek-v4",
      "member": "flash-0731~prem-research-abliterated",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4-Flash-0731",
      "hub_repo": "deepseek-ai/DeepSeek-V4-Flash-0731",
      "publisher": "prem-research",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 159155,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4-Flash-0731",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4-flash-0731",
        "ids": [
          "deepseek/deepseek-v4-flash-0731",
          "deepseek/deepseek-v4-flash-0731:batch"
        ],
        "usd_per_m_in": 0.0434,
        "usd_per_m_out": 0.3,
        "usd_per_m_cached_in": 0.035,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.144,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3.17,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 76.0405,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "propelhq/Kimi-K2.7-Code-mlx-DQ3_K_M-q8",
      "family": "kimi-k2-7",
      "member": "code~propelhq-dq3-k-m",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.7-Code",
      "hub_repo": "moonshotai/Kimi-K2.7-Code",
      "publisher": "propelhq",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 448826,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.7-Code",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.7-code",
        "ids": [
          "moonshotai/kimi-k2.7-code"
        ],
        "usd_per_m_in": 0.7125,
        "usd_per_m_out": 3.5,
        "usd_per_m_cached_in": 0.16,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "pyannote/wespeaker-voxceleb-resnet34-LM",
      "family": "wespeaker",
      "member": "resnet34-lm",
      "quant": "FP32",
      "engine": "speaker-id",
      "modality": "speaker",
      "base": "pyannote/wespeaker-voxceleb-resnet34-LM",
      "hub_repo": "pyannote/wespeaker-voxceleb-resnet34-LM",
      "publisher": "pyannote",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 25,
      "active_share": 1,
      "licence": {
        "id": "cc-by-4.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/pyannote/wespeaker-voxceleb-resnet34-LM",
        "fetched_at": "2026-11-10T15:57:15.731Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T23:42:30.063Z",
          "last_at": "2026-11-14T19:52:48.531Z",
          "caps": [
            "voice.identify"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T19:01:07.978Z",
          "last_at": "2026-11-06T16:52:02.048Z",
          "caps": [
            "voice.identify"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "qwen2.5:0.5b",
      "family": "qwen-2-5",
      "member": "0.5b",
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": null,
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W31"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W31"
          ],
          "tokens_per_sec_best": 4.41,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-07-29T16:45:23.078Z",
          "last_at": "2026-07-29T16:45:23.078Z",
          "caps": [
            "structured_output",
            "tools",
            "tools_loop_terminated"
          ],
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "rhasspy/piper-voices",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "rhasspy/piper-voices",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W34",
        "2026-W37",
        "2026-W39",
        "2026-W44",
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "unknown": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W34"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-23T12:28:38.406Z",
          "last_at": "2026-08-23T12:28:38.406Z",
          "caps": [
            "audio"
          ],
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37",
            "2026-W39",
            "2026-W44",
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-13T07:05:16.558Z",
          "last_at": "2026-11-04T00:40:01.276Z",
          "caps": [
            "audio",
            "audio.out",
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "richardyoung/Kimi-K2-Instruct-0905-MLX-4bit",
      "family": "kimi-k2",
      "member": "instruct-0905",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2-Instruct-0905",
      "hub_repo": "moonshotai/Kimi-K2-Instruct-0905",
      "publisher": "richardyoung",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 550836,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2-Instruct-0905",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2-0905",
        "ids": [
          "moonshotai/kimi-k2-0905"
        ],
        "usd_per_m_in": 0.6,
        "usd_per_m_out": 2.5,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "roman220220/Muse-Glimmer-30B-VL-gptq-mlx-jang",
      "family": "muse-glimmer",
      "member": "30b~roman220220-vl-jang",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "meta-models/Muse-Glimmer-30B",
      "hub_repo": "meta-models/Muse-Glimmer-30B",
      "publisher": "roman220220",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 20484,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/meta-models/Muse-Glimmer-30B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta/muse-glimmer-30b",
        "ids": [
          "meta/muse-glimmer-30b",
          "meta/muse-glimmer-30b:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.04,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.96,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "sarvamai/sarvam-1",
      "family": "sarvam",
      "member": "1",
      "quant": "BF16",
      "engine": "vllm",
      "modality": "text",
      "base": "sarvamai/sarvam-1",
      "hub_repo": "sarvamai/sarvam-1",
      "publisher": "sarvamai",
      "derivative": false,
      "cohort": 1,
      "rung": 3,
      "disk_mib": 4826,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/sarvamai/sarvam-1",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": 1.913,
          "first_at": "2026-10-20T09:39:55.520Z",
          "last_at": "2026-10-20T09:39:55.520Z",
          "caps": [
            "lang.bn",
            "lang.gu",
            "lang.hi",
            "lang.mr",
            "lang.pa",
            "output.long",
            "output.schema",
            "structured_output"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 104.51,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.1047,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "sarvamai/sarvam-translate",
      "family": "sarvam-translate",
      "member": "4b",
      "quant": "BF16",
      "engine": "translate-hf",
      "modality": "translation",
      "base": "google/gemma-3-4b-it",
      "hub_repo": "google/gemma-3-4b-it",
      "publisher": "sarvamai",
      "derivative": false,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 8239,
      "active_share": 1,
      "licence": {
        "id": "gemma",
        "name": null,
        "class": "open-weights",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/google/gemma-3-4b-it",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "google/gemma-3-4b-it",
        "ids": [
          "google/gemma-3-4b-it"
        ],
        "usd_per_m_in": 0.05,
        "usd_per_m_out": 0.1,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-27T00:37:10.261Z",
          "last_at": "2026-10-27T14:26:11.962Z",
          "caps": [
            "translate.as",
            "translate.bn",
            "translate.gu",
            "translate.hi",
            "translate.kn",
            "translate.ml",
            "translate.mr",
            "translate.or",
            "translate.pa",
            "translate.ta",
            "translate.te",
            "translate.text",
            "translate.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T22:04:32.360Z",
          "last_at": "2026-11-06T18:04:03.440Z",
          "caps": [
            "translate.as",
            "translate.bn",
            "translate.gu",
            "translate.hi",
            "translate.kn",
            "translate.ml",
            "translate.mr",
            "translate.ne",
            "translate.or",
            "translate.pa",
            "translate.ta",
            "translate.te",
            "translate.text",
            "translate.ur"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "proved"
        }
      }
    },
    {
      "ref": "segmind/tiny-sd",
      "family": "tiny-sd",
      "member": "base",
      "quant": "FP32",
      "engine": "diffusers",
      "modality": "image",
      "base": null,
      "hub_repo": "segmind/tiny-sd",
      "publisher": "segmind",
      "derivative": false,
      "cohort": 3,
      "rung": 1,
      "disk_mib": 1011,
      "active_share": 1,
      "licence": {
        "id": "creativeml-openrail-m",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/segmind/tiny-sd",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W35",
        "2026-W36",
        "2026-W38",
        "2026-W40",
        "2026-W42",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W40",
            "2026-W42"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-01T07:00:19.297Z",
          "last_at": "2026-10-16T07:37:46.439Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T06:19:16.967Z",
          "last_at": "2026-11-14T13:38:52.239Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T06:31:10.503Z",
          "last_at": "2026-10-22T06:08:22.501Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T21:25:14.031Z",
          "last_at": "2026-11-03T21:25:14.031Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T20:36:46.345Z",
          "last_at": "2026-11-13T23:00:24.897Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W35",
            "2026-W36"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-08-28T18:49:33.806Z",
          "last_at": "2026-09-05T14:11:19.723Z",
          "caps": [
            "image.generate"
          ],
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-14T03:57:27.249Z",
          "last_at": "2026-09-14T03:57:27.249Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "sgl-project/DeepSeek-V4-Flash-FP8",
      "family": "deepseek-v4",
      "member": "flash",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V4-Flash",
      "hub_repo": "deepseek-ai/DeepSeek-V4-Flash",
      "publisher": "sgl-project",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 280417,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V4-Flash",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v4-flash",
        "ids": [
          "deepseek/deepseek-v4-flash"
        ],
        "usd_per_m_in": 0.09,
        "usd_per_m_out": 0.18,
        "usd_per_m_cached_in": 0.018,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.8,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 188.7398,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "sh0wie/Qwen3.8-Flash-Next-REAP-288-MLX-4bit",
      "family": "qwen-3-8",
      "member": "flash-next~sh0wie-reap-288",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "Qwen/Qwen3.8-Flash-Next",
      "hub_repo": "Qwen/Qwen3.8-Flash-Next",
      "publisher": "sh0wie",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 70087,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "qwen-community-1.0",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-Flash-Next",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "qwen/qwen3.8-flash",
        "ids": [
          "qwen/qwen3.8-flash"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.47,
        "usd_per_m_cached_in": 0.016,
        "context_length": 1000000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "sjug/DeepSeek-V3.2-MLX-4bit",
      "family": "deepseek-v3-2",
      "member": "base~sjug",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V3.2",
      "hub_repo": "deepseek-ai/DeepSeek-V3.2",
      "publisher": "sjug",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 360571,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.2",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v3.2",
        "ids": [
          "deepseek/deepseek-v3.2"
        ],
        "usd_per_m_in": 0.26,
        "usd_per_m_out": 0.38,
        "usd_per_m_cached_in": 0.13,
        "context_length": 163840,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "sombra-x/Kimi-K2.6-REAP-Solidity",
      "family": "kimi-k2-6",
      "member": "base~sombra-x-reap-solidity",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.6",
      "hub_repo": "moonshotai/Kimi-K2.6",
      "publisher": "sombra-x",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 410300,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.6",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.6",
        "ids": [
          "moonshotai/kimi-k2.6"
        ],
        "usd_per_m_in": 0.65,
        "usd_per_m_out": 3.41,
        "usd_per_m_cached_in": 0.15,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "speechbrain/spkrec-ecapa-voxceleb",
      "family": "ecapa-tdnn",
      "member": "voxceleb",
      "quant": "FP32",
      "engine": "speaker-id",
      "modality": "speaker",
      "base": "speechbrain/spkrec-ecapa-voxceleb",
      "hub_repo": "speechbrain/spkrec-ecapa-voxceleb",
      "publisher": "speechbrain",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 85,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/speechbrain/spkrec-ecapa-voxceleb",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43",
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-25T23:43:43.739Z",
          "last_at": "2026-11-14T20:02:52.174Z",
          "caps": [
            "voice.identify"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T19:19:52.962Z",
          "last_at": "2026-11-06T17:02:07.301Z",
          "caps": [
            "voice.identify"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "spicyneuron/GLM-5.1-MLX-4.5bit",
      "family": "glm-5-1",
      "member": "base~spicyneuron-4",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-5.1",
      "hub_repo": "zai-org/GLM-5.1",
      "publisher": "spicyneuron",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 402424,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.1",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.1",
        "ids": [
          "z-ai/glm-5.1"
        ],
        "usd_per_m_in": 1.4,
        "usd_per_m_out": 4.4,
        "usd_per_m_cached_in": 0.26,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "spicyneuron/Kimi-K2.5-MLX-2.8bit",
      "family": "kimi-k2-5",
      "member": "base~spicyneuron-2",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.5",
      "hub_repo": "moonshotai/Kimi-K2.5",
      "publisher": "spicyneuron",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 351246,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.5",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.5",
        "ids": [
          "moonshotai/kimi-k2.5"
        ],
        "usd_per_m_in": 0.5,
        "usd_per_m_out": 2.5,
        "usd_per_m_cached_in": 0.15,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "spicyneuron/Kimi-K2.6-MLX-3.3bit",
      "family": "kimi-k2-6",
      "member": "base~spicyneuron-3",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.6",
      "hub_repo": "moonshotai/Kimi-K2.6",
      "publisher": "spicyneuron",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 407601,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.6",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.6",
        "ids": [
          "moonshotai/kimi-k2.6"
        ],
        "usd_per_m_in": 0.65,
        "usd_per_m_out": 3.41,
        "usd_per_m_cached_in": 0.15,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "spicyneuron/Kimi-K2.7-Code-MLX-3.6bit",
      "family": "kimi-k2-7",
      "member": "code~spicyneuron-3",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.7-Code",
      "hub_repo": "moonshotai/Kimi-K2.7-Code",
      "publisher": "spicyneuron",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 437841,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.7-Code",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.7-code",
        "ids": [
          "moonshotai/kimi-k2.7-code"
        ],
        "usd_per_m_in": 0.7125,
        "usd_per_m_out": 3.5,
        "usd_per_m_cached_in": 0.16,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "stabilityai/sd-turbo",
      "family": "sd-turbo",
      "member": "2-1",
      "quant": "FP16",
      "engine": "diffusers",
      "modality": "image",
      "base": null,
      "hub_repo": "stabilityai/sd-turbo",
      "publisher": "stabilityai",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 2460,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/stabilityai/sd-turbo",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W37",
        "2026-W38",
        "2026-W43",
        "2026-W44",
        "2026-W45",
        "2026-W46"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 2,
          "seasons": [
            "2026-W44",
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T06:18:09.926Z",
          "last_at": "2026-11-11T06:16:14.890Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T07:04:12.609Z",
          "last_at": "2026-10-22T08:16:35.492Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-03T21:24:17.998Z",
          "last_at": "2026-11-03T21:24:17.998Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-15T21:07:30.512Z",
          "last_at": "2026-09-15T21:07:30.512Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/none/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W37"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-08T19:10:15.129Z",
          "last_at": "2026-09-08T19:10:15.129Z",
          "caps": [
            "image.generate"
          ],
          "verdict": "only the network runs it"
        },
        "windows/amd64/nvidia/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W38"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-09-14T04:05:59.345Z",
          "last_at": "2026-09-14T04:05:59.345Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "staccs/Muse-Spark-Open-816B-A42B-lecore-mlx",
      "family": "muse-spark",
      "member": "816b-a42b~staccs-open-lecore",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "meta-models/Muse-Spark-Open-816B-A42B",
      "hub_repo": "meta-models/Muse-Spark-Open-816B-A42B",
      "publisher": "staccs",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 248203,
      "active_share": 0.0515,
      "licence": {
        "id": "other",
        "name": "muse-model-license",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": true,
        "source": "https://huggingface.co/api/models/meta-models/Muse-Spark-Open-816B-A42B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "meta/muse-spark-open",
        "ids": [
          "meta/muse-spark-open"
        ],
        "usd_per_m_in": 1.25,
        "usd_per_m_out": 4.25,
        "usd_per_m_cached_in": 0.15,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "stepfun-ai/Step-3.7-Flash-FP8",
      "family": "step-3-7",
      "member": "flash",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "stepfun-ai/Step-3.7-Flash-FP8",
      "hub_repo": "stepfun-ai/Step-3.7-Flash-FP8",
      "publisher": "stepfun-ai",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 202678,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash-FP8",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "stepfun/step-3.7-flash",
        "ids": [
          "stepfun/step-3.7-flash"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 1.15,
        "usd_per_m_cached_in": 0.04,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 111.0311,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "tebruno99/Mistral-Medium-3.5-128B-4bit",
      "family": "mistral-medium-3-5",
      "member": "128b~tebruno99",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": null,
      "hub_repo": "tebruno99/Mistral-Medium-3.5-128B-4bit",
      "publisher": "tebruno99",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 74388,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Medium-3.5-128B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-medium-3-5",
        "ids": [
          "mistralai/mistral-medium-3-5",
          "mistralai/mistral-medium-3-5:batch"
        ],
        "usd_per_m_in": 1.5,
        "usd_per_m_out": 7.5,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": 3.75,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "tencent/Hy-MT2-1.8B-FP8",
      "family": "hy-mt-2",
      "member": "1.8b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "tencent/Hy-MT2-1.8B",
      "hub_repo": "tencent/Hy-MT2-1.8B",
      "publisher": "tencent",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 1944,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy-MT2-1.8B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy-mt2-1.8b",
        "ids": [
          "tencent/hy-mt2-1.8b"
        ],
        "usd_per_m_in": 0.044,
        "usd_per_m_out": 0.177,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 259.44,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.436,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "tencent/Hy-MT2-30B-A3B-FP8",
      "family": "hy-mt-2",
      "member": "30b-a3b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "tencent/Hy-MT2-30B-A3B",
      "hub_repo": "tencent/Hy-MT2-30B-A3B",
      "publisher": "tencent",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 29162,
      "active_share": 0.1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy-MT2-30B-A3B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy-mt2-30b-a3b",
        "ids": [
          "tencent/hy-mt2-30b-a3b"
        ],
        "usd_per_m_in": 0.074,
        "usd_per_m_out": 0.295,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 172.95,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7821,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "tencent/Hy-MT2-7B-FP8",
      "family": "hy-mt-2",
      "member": "7b",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "tencent/Hy-MT2-7B",
      "hub_repo": "tencent/Hy-MT2-7B",
      "publisher": "tencent",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 7657,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy-MT2-7B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy-mt2-7b",
        "ids": [
          "tencent/hy-mt2-7b"
        ],
        "usd_per_m_in": 0.074,
        "usd_per_m_out": 0.295,
        "usd_per_m_cached_in": null,
        "context_length": 8192,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 65.87,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 1.7877,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "tencent/Hy3-FP8",
      "family": "hy-3",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "tencent/Hy3-FP8",
      "hub_repo": "tencent/Hy3-FP8",
      "publisher": "tencent",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 285997,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy3-FP8",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "tencent/hy3",
        "ids": [
          "tencent/hy3"
        ],
        "usd_per_m_in": 0.0825,
        "usd_per_m_out": 0.33,
        "usd_per_m_cached_in": 0.020625,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 195.6095,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "terion-mlx/GLM-4.5-Air-mixed_4_6",
      "family": "glm-4-5",
      "member": "air~terion-mlx-mixed-4-6",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "zai-org/GLM-4.5-Air",
      "hub_repo": "zai-org/GLM-4.5-Air",
      "publisher": "terion-mlx",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 61259,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.5-Air",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.5-air",
        "ids": [
          "z-ai/glm-4.5-air"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 1.1,
        "usd_per_m_cached_in": 0.03,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "thetom-ai/MiniMax-M2.7-ConfigI-MLX",
      "family": "minimax-m2-7",
      "member": "base~thetom-ai-configi",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.7",
      "hub_repo": "MiniMaxAI/MiniMax-M2.7",
      "publisher": "thetom-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 88569,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "other",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.7",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.7",
        "ids": [
          "minimax/minimax-m2.7"
        ],
        "usd_per_m_in": 0.21,
        "usd_per_m_out": 0.84,
        "usd_per_m_cached_in": 0.042,
        "context_length": 196608,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "thetom-ai/MiniMax-M3-ConfigI-MLX",
      "family": "minimax-m3",
      "member": "base~thetom-ai-configi",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M3",
      "hub_repo": "MiniMaxAI/MiniMax-M3",
      "publisher": "thetom-ai",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 159333,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "minimax-community",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M3",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m3",
        "ids": [
          "minimax/minimax-m3",
          "minimax/minimax-m3:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.06,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.88,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unigilby/MiniMax-M3-oQ4",
      "family": "minimax-m3",
      "member": "base~unigilby-oq4",
      "quant": "4BIT",
      "engine": "mlx",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M3",
      "hub_repo": "MiniMaxAI/MiniMax-M3",
      "publisher": "unigilby",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 233624,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "minimax-community",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M3",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m3",
        "ids": [
          "minimax/minimax-m3",
          "minimax/minimax-m3:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.06,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.88,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/DeepSeek-V3.1-Terminus-GGUF:Q4_K_M",
      "family": "deepseek-v3-1",
      "member": "terminus",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V3.1-Terminus",
      "hub_repo": "deepseek-ai/DeepSeek-V3.1-Terminus",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 386617,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.1-Terminus",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v3.1-terminus",
        "ids": [
          "deepseek/deepseek-v3.1-terminus"
        ],
        "usd_per_m_in": 0.27,
        "usd_per_m_out": 1,
        "usd_per_m_cached_in": null,
        "context_length": 163840,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/DeepSeek-V3.2-GGUF:Q4_K_M",
      "family": "deepseek-v3-2",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "deepseek-ai/DeepSeek-V3.2",
      "hub_repo": "deepseek-ai/DeepSeek-V3.2",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 386617,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/deepseek-ai/DeepSeek-V3.2",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "deepseek/deepseek-v3.2",
        "ids": [
          "deepseek/deepseek-v3.2"
        ],
        "usd_per_m_in": 0.26,
        "usd_per_m_out": 0.38,
        "usd_per_m_cached_in": 0.13,
        "context_length": 163840,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/Devstral-2-123B-Instruct-2512-GGUF:Q4_K_M",
      "family": "devstral-2",
      "member": "123b-instruct-2512",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "mistralai/Devstral-2-123B-Instruct-2512",
      "hub_repo": "mistralai/Devstral-2-123B-Instruct-2512",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 71427,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Devstral-2-123B-Instruct-2512",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/devstral-2512",
        "ids": [
          "mistralai/devstral-2512"
        ],
        "usd_per_m_in": 0.4,
        "usd_per_m_out": 2,
        "usd_per_m_cached_in": 0.04,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/GLM-4.5-Air-GGUF:Q4_K_M",
      "family": "glm-4-5",
      "member": "air",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "zai-org/GLM-4.5-Air",
      "hub_repo": "zai-org/GLM-4.5-Air",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 69595,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.5-Air",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.5-air",
        "ids": [
          "z-ai/glm-4.5-air"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 1.1,
        "usd_per_m_cached_in": 0.03,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/GLM-4.5-GGUF:Q4_K_M",
      "family": "glm-4-5",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "zai-org/GLM-4.5",
      "hub_repo": "zai-org/GLM-4.5",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 206428,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.5",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.5",
        "ids": [
          "z-ai/glm-4.5"
        ],
        "usd_per_m_in": 0.6,
        "usd_per_m_out": 2.2,
        "usd_per_m_cached_in": 0.11,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/GLM-4.6-GGUF:Q4_K_M",
      "family": "glm-4-6",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "zai-org/GLM-4.6",
      "hub_repo": "zai-org/GLM-4.6",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 205595,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.6",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.6",
        "ids": [
          "z-ai/glm-4.6"
        ],
        "usd_per_m_in": 0.5,
        "usd_per_m_out": 2,
        "usd_per_m_cached_in": 0.1,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/GLM-4.6V-GGUF:Q4_K_M",
      "family": "glm-4-6",
      "member": "v",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "zai-org/GLM-4.6V",
      "hub_repo": "zai-org/GLM-4.6V",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 67296,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.6V",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.6v",
        "ids": [
          "z-ai/glm-4.6v"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.055,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/GLM-4.7-Flash-FP8-Dynamic",
      "family": "glm-4-7",
      "member": "flash",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-4.7-Flash",
      "hub_repo": "zai-org/GLM-4.7-Flash",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 31015,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.7-Flash",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.7-flash",
        "ids": [
          "z-ai/glm-4.7-flash"
        ],
        "usd_per_m_in": 0.0605,
        "usd_per_m_out": 0.4,
        "usd_per_m_cached_in": null,
        "context_length": 200000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.26,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.4113,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "unsloth/GLM-4.7-GGUF:Q4_K_M",
      "family": "glm-4-7",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "zai-org/GLM-4.7",
      "hub_repo": "zai-org/GLM-4.7",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 206428,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.7",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.7",
        "ids": [
          "z-ai/glm-4.7"
        ],
        "usd_per_m_in": 0.4,
        "usd_per_m_out": 1.75,
        "usd_per_m_cached_in": 0.08,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/GLM-5-GGUF:Q4_K_M",
      "family": "glm-5",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "zai-org/GLM-5",
      "hub_repo": "zai-org/GLM-5",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 434760,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5",
        "ids": [
          "z-ai/glm-5"
        ],
        "usd_per_m_in": 1,
        "usd_per_m_out": 3.2,
        "usd_per_m_cached_in": 0.2,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/GLM-5.1-GGUF:Q4_K_M",
      "family": "glm-5-1",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "zai-org/GLM-5.1",
      "hub_repo": "zai-org/GLM-5.1",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 442985,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.1",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.1",
        "ids": [
          "z-ai/glm-5.1"
        ],
        "usd_per_m_in": 1.4,
        "usd_per_m_out": 4.4,
        "usd_per_m_cached_in": 0.26,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/GLM-5.2-GGUF:Q4_K_M",
      "family": "glm-5-2",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "zai-org/GLM-5.2",
      "hub_repo": "zai-org/GLM-5.2",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 444245,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.2",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.2",
        "ids": [
          "z-ai/glm-5.2"
        ],
        "usd_per_m_in": 0.007,
        "usd_per_m_out": 6.8,
        "usd_per_m_cached_in": 0.006,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/Kimi-K2-Instruct-0905-GGUF:Q4_K_M",
      "family": "kimi-k2",
      "member": "instruct-0905",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "moonshotai/Kimi-K2-Instruct-0905",
      "hub_repo": "moonshotai/Kimi-K2-Instruct-0905",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 592465,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2-Instruct-0905",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2-0905",
        "ids": [
          "moonshotai/kimi-k2-0905"
        ],
        "usd_per_m_in": 0.6,
        "usd_per_m_out": 2.5,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/Kimi-K2-Instruct-GGUF:Q4_K_M",
      "family": "kimi-k2",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "moonshotai/Kimi-K2-Instruct",
      "hub_repo": "moonshotai/Kimi-K2-Instruct",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 592023,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2-Instruct",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2",
        "ids": [
          "moonshotai/kimi-k2"
        ],
        "usd_per_m_in": 0.57,
        "usd_per_m_out": 2.3,
        "usd_per_m_cached_in": null,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/Kimi-K2-Thinking-GGUF:Q4_K_M",
      "family": "kimi-k2",
      "member": "thinking",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "moonshotai/Kimi-K2-Thinking",
      "hub_repo": "moonshotai/Kimi-K2-Thinking",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 592465,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2-Thinking",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2-thinking",
        "ids": [
          "moonshotai/kimi-k2-thinking"
        ],
        "usd_per_m_in": 0.6,
        "usd_per_m_out": 2.5,
        "usd_per_m_cached_in": 0.15,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/Kimi-K2.5-GGUF:Q4_K_M",
      "family": "kimi-k2-5",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "moonshotai/Kimi-K2.5",
      "hub_repo": "moonshotai/Kimi-K2.5",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 592465,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/moonshotai/Kimi-K2.5",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "moonshotai/kimi-k2.5",
        "ids": [
          "moonshotai/kimi-k2.5"
        ],
        "usd_per_m_in": 0.5,
        "usd_per_m_out": 2.5,
        "usd_per_m_cached_in": 0.15,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/MiMo-V2.5-GGUF:Q4_K_M",
      "family": "mimo-v2-5",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "XiaomiMiMo/MiMo-V2.5",
      "hub_repo": "XiaomiMiMo/MiMo-V2.5",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 182072,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.5",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.5",
        "ids": [
          "xiaomi/mimo-v2.5"
        ],
        "usd_per_m_in": 0.4,
        "usd_per_m_out": 2,
        "usd_per_m_cached_in": 0.08,
        "context_length": 1000000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/MiMo-V2.5-Pro-GGUF:Q4_K_M",
      "family": "mimo-v2-5",
      "member": "pro",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "XiaomiMiMo/MiMo-V2.5-Pro",
      "hub_repo": "XiaomiMiMo/MiMo-V2.5-Pro",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 600445,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/XiaomiMiMo/MiMo-V2.5-Pro",
        "fetched_at": "2026-11-10T14:41:51.909Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "xiaomi/mimo-v2.5-pro",
        "ids": [
          "xiaomi/mimo-v2.5-pro"
        ],
        "usd_per_m_in": 0.435,
        "usd_per_m_out": 0.87,
        "usd_per_m_cached_in": 0.0036,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/MiniMax-M2.5-GGUF:Q4_K_M",
      "family": "minimax-m2-5",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.5",
      "hub_repo": "MiniMaxAI/MiniMax-M2.5",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 131933,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "modified-mit",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.5",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.5",
        "ids": [
          "minimax/minimax-m2.5"
        ],
        "usd_per_m_in": 0.27,
        "usd_per_m_out": 0.95,
        "usd_per_m_cached_in": 0.03,
        "context_length": 198000,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/MiniMax-M2.7-GGUF:Q4_K_M",
      "family": "minimax-m2-7",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M2.7",
      "hub_repo": "MiniMaxAI/MiniMax-M2.7",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 133676,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "other",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M2.7",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m2.7",
        "ids": [
          "minimax/minimax-m2.7"
        ],
        "usd_per_m_in": 0.21,
        "usd_per_m_out": 0.84,
        "usd_per_m_cached_in": 0.042,
        "context_length": 196608,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/MiniMax-M3-GGUF:Q4_K_M",
      "family": "minimax-m3",
      "member": "base",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M3",
      "hub_repo": "MiniMaxAI/MiniMax-M3",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 252655,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "minimax-community",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M3",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m3",
        "ids": [
          "minimax/minimax-m3",
          "minimax/minimax-m3:batch"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 1.2,
        "usd_per_m_cached_in": 0.06,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.88,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/Mistral-Large-3-675B-Instruct-2512-GGUF:Q4_K_M",
      "family": "mistral-large-3",
      "member": "675b-instruct-2512",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "mistralai/Mistral-Large-3-675B-Instruct-2512",
      "hub_repo": "mistralai/Mistral-Large-3-675B-Instruct-2512",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 388135,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Large-3-675B-Instruct-2512",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-large-2512",
        "ids": [
          "mistralai/mistral-large-2512"
        ],
        "usd_per_m_in": 0.5,
        "usd_per_m_out": 1.5,
        "usd_per_m_cached_in": 0.05,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/Mistral-Medium-3.5-128B-GGUF:Q4_K_M",
      "family": "mistral-medium-3-5",
      "member": "128b",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "mistralai/Mistral-Medium-3.5-128B",
      "hub_repo": "mistralai/Mistral-Medium-3.5-128B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 71427,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Medium-3.5-128B",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-medium-3-5",
        "ids": [
          "mistralai/mistral-medium-3-5",
          "mistralai/mistral-medium-3-5:batch"
        ],
        "usd_per_m_in": 1.5,
        "usd_per_m_out": 7.5,
        "usd_per_m_cached_in": null,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": 3.75,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/Mistral-Small-4-119B-2603-GGUF:Q4_K_M",
      "family": "mistral-small-4",
      "member": "119b-2603",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "mistralai/Mistral-Small-4-119B-2603",
      "hub_repo": "mistralai/Mistral-Small-4-119B-2603",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 70346,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/mistralai/Mistral-Small-4-119B-2603",
        "fetched_at": "2026-11-10T15:12:00.247Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "mistralai/mistral-small-2603",
        "ids": [
          "mistralai/mistral-small-2603"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": 0.015,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/Qwen3.5-35B-A3B-GGUF:IQ2_M",
      "family": "qwen-3-5",
      "member": "35b-a3b",
      "quant": "IQ2_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.5-35B-A3B",
      "hub_repo": "Qwen/Qwen3.5-35B-A3B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 9,
      "disk_mib": 10163,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.5-35B-A3B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W45",
        "2026-W46"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.5-35b-a3b",
        "ids": [
          "qwen/qwen3.5-35b-a3b"
        ],
        "usd_per_m_in": 0.14,
        "usd_per_m_out": 1,
        "usd_per_m_cached_in": 0.05,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 235.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3361,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45",
            "2026-W46"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 11.516,
          "first_at": "2026-11-05T16:38:05.710Z",
          "last_at": "2026-11-14T09:28:40.233Z",
          "caps": [
            "code.reason",
            "code.write",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call"
          ],
          "est_tokens_per_sec": 54.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7114,
          "tariff_basis": "bootstrap",
          "verdict": "cheaper than the API"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1564.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3098,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 541.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0714,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 219.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.546,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.8639,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.1523,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 794.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3329,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "unsloth/Qwen3.5-35B-A3B-GGUF:IQ2_XXS",
      "family": "qwen-3-5",
      "member": "35b-a3b",
      "quant": "IQ2_XXS",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.5-35B-A3B",
      "hub_repo": "Qwen/Qwen3.5-35B-A3B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 9,
      "disk_mib": 10163,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.5-35B-A3B",
        "fetched_at": "2026-11-10T15:27:06.468Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.5-35b-a3b",
        "ids": [
          "qwen/qwen3.5-35b-a3b"
        ],
        "usd_per_m_in": 0.14,
        "usd_per_m_out": 1,
        "usd_per_m_cached_in": 0.05,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 235.76,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3361,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": 9.312,
          "first_at": "2026-11-05T01:52:16.623Z",
          "last_at": "2026-11-05T01:52:16.623Z",
          "caps": [
            "code.reason",
            "code.write",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.kn",
            "lang.mr",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "structured_output",
            "thinking",
            "tools",
            "tools.call"
          ],
          "est_tokens_per_sec": 54.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.7114,
          "tariff_basis": "bootstrap",
          "verdict": "cheaper than the API"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1564.84,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3098,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 541.48,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0714,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 219.42,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.546,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.94,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.8639,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 15.38,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 5.1523,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 794.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3329,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "unsloth/Qwen3.6-35B-A3B-GGUF:IQ1_M",
      "family": "qwen-3-6",
      "member": "35b-a3b",
      "quant": "IQ1_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.6-35B-A3B",
      "hub_repo": "Qwen/Qwen3.6-35B-A3B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 5,
      "disk_mib": 9582,
      "active_share": 0.0857,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.6-35B-A3B",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.6-35b-a3b",
        "ids": [
          "qwen/qwen3.6-35b-a3b"
        ],
        "usd_per_m_in": 0.1,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.025,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 250.05,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.315,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 8.563,
          "first_at": "2026-11-08T03:28:28.486Z",
          "last_at": "2026-11-08T03:28:28.486Z",
          "caps": [
            "code.reason",
            "code.write",
            "faithful_tool_result",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call",
            "tools.faithful",
            "tools.multi_hop"
          ],
          "est_tokens_per_sec": 57.68,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.6625,
          "tariff_basis": "bootstrap",
          "verdict": "cheaper than the API"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1659.72,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.2918,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 574.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0665,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 232.73,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.5127,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 9.49,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 8.3003,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 16.31,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.8296,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 842.79,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.3133,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "unsloth/Qwen3.8-27B-GGUF:Q4_K_M",
      "family": "qwen-3-8",
      "member": "27b",
      "quant": "Q4_K_M",
      "engine": "ollama",
      "modality": "text",
      "base": "Qwen/Qwen3.8-27B",
      "hub_repo": "Qwen/Qwen3.8-27B",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 1,
      "rung": 5,
      "disk_mib": 5905,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/Qwen/Qwen3.8-27B",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [
        "2026-W44",
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "qwen/qwen3.8-27b",
        "ids": [
          "qwen/qwen3.8-27b",
          "qwen/qwen3.8-27b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 1.875,
        "usd_per_m_cached_in": 0.0375,
        "context_length": 1024000,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "darwin/arm64/apple/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 34.78,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.1788,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/none/32g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 8.02,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 4.3918,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/256g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 230.85,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.085,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": 8192,
          "joules_per_token_best": 11.33,
          "first_at": "2026-10-28T07:01:10.429Z",
          "last_at": "2026-10-28T07:01:10.429Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call"
          ],
          "est_tokens_per_sec": 79.88,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.4409,
          "tariff_basis": "bootstrap",
          "verdict": "cheaper than the API"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": 4096,
          "joules_per_token_best": 41.561,
          "first_at": "2026-11-07T01:35:22.479Z",
          "last_at": "2026-11-07T01:35:22.479Z",
          "caps": [
            "code.reason",
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.gu",
            "lang.hi",
            "lang.kn",
            "lang.ml",
            "lang.mr",
            "lang.pa",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call"
          ],
          "est_tokens_per_sec": 32.37,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 3.5939,
          "tariff_basis": "bootstrap",
          "verdict": "the API is cheaper"
        },
        "windows/amd64/amd/8g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.32,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 57.4074,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        },
        "windows/amd64/intel/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 2.27,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 33.3823,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        },
        "windows/amd64/nvidia/16g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 117.22,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 2.2271,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "unsloth/Step-3.7-Flash-GGUF:Q4_K_M",
      "family": "step-3-7",
      "member": "flash",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "stepfun-ai/Step-3.7-Flash",
      "hub_repo": "stepfun-ai/Step-3.7-Flash",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 116434,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash",
        "fetched_at": "2026-11-10T15:42:11.362Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "stepfun/step-3.7-flash",
        "ids": [
          "stepfun/step-3.7-flash"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 1.15,
        "usd_per_m_cached_in": 0.04,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "unsloth/gemma-4-E2B-it-qat-mobile-GGUF:Q2_K_XL",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "unsloth/gemma-4-E2B-it-qat-mobile-GGUF",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": null,
        "name": null,
        "class": "unknown",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": null,
        "fetched_at": null
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": 3.058,
          "first_at": "2026-10-30T03:59:08.665Z",
          "last_at": "2026-10-30T03:59:08.665Z",
          "caps": [
            "code.write",
            "instruct.follow",
            "lang.bn",
            "lang.en",
            "lang.hi",
            "lang.ml",
            "lang.ta",
            "lang.te",
            "output.long",
            "output.schema",
            "reason.channel",
            "research.synthesize",
            "structured_output",
            "thinking",
            "tools",
            "tools.call"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "unsloth/gpt-oss-120b-GGUF:Q4_K_M",
      "family": "gpt-oss",
      "member": "120b",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "openai/gpt-oss-120b",
      "hub_repo": "openai/gpt-oss-120b",
      "publisher": "unsloth",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 59860,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/gpt-oss-120b",
        "fetched_at": "2026-11-10T13:56:03.873Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "openai/gpt-oss-120b",
        "ids": [
          "openai/gpt-oss-120b",
          "openai/gpt-oss-120b:batch"
        ],
        "usd_per_m_in": 0.15,
        "usd_per_m_out": 0.6,
        "usd_per_m_cached_in": 0.075,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": 0.136,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "ununununununununununununununununununun/sd-turbo-phi-lattice",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "ununununununununununununununununununun/sd-turbo-phi-lattice",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ununununununununununununununununununun/sd-turbo-phi-lattice",
        "fetched_at": "2026-11-05T03:34:22.455Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-08T01:14:59.288Z",
          "last_at": "2026-11-08T01:14:59.288Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-06T09:21:26.590Z",
          "last_at": "2026-11-06T09:21:26.590Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-07T20:41:58.319Z",
          "last_at": "2026-11-07T20:41:58.319Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "ununununununununununununununununununun/sd-turbo-phi-lattice-v2",
      "family": null,
      "member": null,
      "quant": null,
      "engine": null,
      "modality": "other",
      "base": null,
      "hub_repo": "ununununununununununununununununununun/sd-turbo-phi-lattice-v2",
      "publisher": null,
      "derivative": false,
      "cohort": null,
      "rung": null,
      "disk_mib": null,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/ununununununununununununununununununun/sd-turbo-phi-lattice-v2",
        "fetched_at": "2026-11-05T03:34:22.455Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-06T09:24:30.962Z",
          "last_at": "2026-11-06T09:24:30.962Z",
          "caps": [
            "image.generate"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        },
        "windows/amd64/intel/16g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-11-06T22:22:41.059Z",
          "last_at": "2026-11-06T22:22:41.059Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "vasista22/whisper-hindi-small",
      "family": "whisper-indic-ft",
      "member": "hi-small",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "openai/whisper-small",
      "hub_repo": "openai/whisper-small",
      "publisher": "vasista22",
      "derivative": false,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 922,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/whisper-small",
        "fetched_at": "2026-11-10T15:57:15.731Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T18:52:26.303Z",
          "last_at": "2026-10-27T11:33:45.752Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.hi"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "vasista22/whisper-kannada-small",
      "family": "whisper-indic-ft",
      "member": "kn-small",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "openai/whisper-small",
      "hub_repo": "openai/whisper-small",
      "publisher": "vasista22",
      "derivative": false,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 922,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/whisper-small",
        "fetched_at": "2026-11-10T15:57:15.731Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-30T18:01:21.705Z",
          "last_at": "2026-10-30T18:01:21.705Z",
          "caps": [
            "audio",
            "audio.in"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "vasista22/whisper-tamil-small",
      "family": "whisper-indic-ft",
      "member": "ta-small",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "openai/whisper-small",
      "hub_repo": "openai/whisper-small",
      "publisher": "vasista22",
      "derivative": false,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 922,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/whisper-small",
        "fetched_at": "2026-11-10T15:57:15.731Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T19:38:00.301Z",
          "last_at": "2026-10-27T12:01:05.273Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.ta"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "vasista22/whisper-telugu-small",
      "family": "whisper-indic-ft",
      "member": "te-small",
      "quant": "FP32",
      "engine": "speech-hf",
      "modality": "speech",
      "base": "openai/whisper-small",
      "hub_repo": "openai/whisper-small",
      "publisher": "vasista22",
      "derivative": false,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 922,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/openai/whisper-small",
        "fetched_at": "2026-11-10T15:57:15.731Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/none/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-26T20:40:08.800Z",
          "last_at": "2026-10-26T21:17:15.583Z",
          "caps": [
            "audio",
            "audio.in",
            "lang.te"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "webmp3/Sakura-MiniMax-M3.1-Flash-Q4_K_M-GGUF:Q4_K_M",
      "family": "minimax-m3-1",
      "member": "flash~webmp3-sakura-k-m",
      "quant": "Q4_K_M",
      "engine": "llamacpp",
      "modality": "text",
      "base": "MiniMaxAI/MiniMax-M3.1-Flash",
      "hub_repo": "MiniMaxAI/MiniMax-M3.1-Flash",
      "publisher": "webmp3",
      "derivative": true,
      "cohort": 3,
      "rung": 0,
      "disk_mib": 246138,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "minimax-community",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/MiniMaxAI/MiniMax-M3.1-Flash",
        "fetched_at": "2026-11-10T14:57:05.380Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "minimax/minimax-m3.1-flash",
        "ids": [
          "minimax/minimax-m3.1-flash"
        ],
        "usd_per_m_in": 0.075,
        "usd_per_m_out": 0.3,
        "usd_per_m_cached_in": 0.0075,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {}
    },
    {
      "ref": "yujiepan/ernie-4.5-moe-tiny-random",
      "family": "ernie-4-5",
      "member": "21b-a3b-thinking~yujiepan-moe-tiny-random",
      "quant": "BF16",
      "engine": "vllm",
      "modality": "text",
      "base": "baidu/ERNIE-4.5-21B-A3B-Thinking",
      "hub_repo": "baidu/ERNIE-4.5-21B-A3B-Thinking",
      "publisher": "yujiepan",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 1,
      "active_share": 0.1429,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/baidu/ERNIE-4.5-21B-A3B-Thinking",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": 0.497,
          "first_at": "2026-10-28T04:12:44.583Z",
          "last_at": "2026-10-28T04:12:44.583Z",
          "caps": [
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 3530466.34,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "yujiepan/ernie-4.5-tiny-random",
      "family": "ernie-4-5",
      "member": "0.3b~yujiepan-tiny-random",
      "quant": "BF16",
      "engine": "vllm",
      "modality": "text",
      "base": "baidu/ERNIE-4.5-0.3B-PT",
      "hub_repo": "baidu/ERNIE-4.5-0.3B-PT",
      "publisher": "yujiepan",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 1,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/baidu/ERNIE-4.5-0.3B-PT",
        "fetched_at": "2026-11-10T13:10:14.835Z"
      },
      "seasons": [
        "2026-W44"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W44"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": 0.471,
          "first_at": "2026-10-28T03:23:19.594Z",
          "last_at": "2026-10-28T03:23:19.594Z",
          "caps": [
            "lang.en"
          ],
          "tariff_basis": "bootstrap",
          "verdict": "no hosted API listed"
        },
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 504352.33,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0002,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "yujiepan/hy3-tiny-random",
      "family": "hy-3",
      "member": "base~yujiepan-tiny-random",
      "quant": "BF16",
      "engine": "vllm",
      "modality": "text",
      "base": "tencent/Hy3",
      "hub_repo": "tencent/Hy3",
      "publisher": "yujiepan",
      "derivative": true,
      "cohort": 3,
      "rung": 3,
      "disk_mib": 5,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/tencent/Hy3",
        "fetched_at": "2026-11-10T14:11:18.411Z"
      },
      "seasons": [
        "2026-W45"
      ],
      "api": {
        "openrouter_id": "tencent/hy3",
        "ids": [
          "tencent/hy3"
        ],
        "usd_per_m_in": 0.0825,
        "usd_per_m_out": 0.33,
        "usd_per_m_cached_in": 0.020625,
        "context_length": 262144,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W45"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": 0.141,
          "first_at": "2026-11-03T21:13:26.540Z",
          "last_at": "2026-11-03T21:13:26.540Z",
          "caps": [],
          "est_tokens_per_sec": 100870.47,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 0.0011,
          "tariff_basis": "bootstrap",
          "verdict": "cheaper than the API"
        }
      }
    },
    {
      "ref": "zai-org/CogVideoX-2b",
      "family": "cogvideox",
      "member": "2b",
      "quant": null,
      "engine": "diffusers",
      "modality": "image",
      "base": null,
      "hub_repo": "zai-org/CogVideoX-2b",
      "publisher": "zai-org",
      "derivative": false,
      "cohort": 3,
      "rung": 9,
      "disk_mib": 13136,
      "active_share": 1,
      "licence": {
        "id": "apache-2.0",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/CogVideoX-2b",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-21T06:34:26.383Z",
          "last_at": "2026-10-21T06:34:26.383Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "zai-org/CogVideoX-5b",
      "family": "cogvideox",
      "member": "5b",
      "quant": null,
      "engine": "diffusers",
      "modality": "image",
      "base": null,
      "hub_repo": "zai-org/CogVideoX-5b",
      "publisher": "zai-org",
      "derivative": false,
      "cohort": 3,
      "rung": 9,
      "disk_mib": 20530,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": null,
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/CogVideoX-5b",
        "fetched_at": "2026-11-10T12:54:59.674Z"
      },
      "seasons": [
        "2026-W43"
      ],
      "api": null,
      "no_hosted_api": true,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/32g": {
          "proved": true,
          "grams": 1,
          "seasons": [
            "2026-W43"
          ],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": "2026-10-23T20:12:21.561Z",
          "last_at": "2026-10-23T20:12:21.561Z",
          "caps": [],
          "tariff_basis": "bootstrap",
          "verdict": "only the network runs it"
        }
      }
    },
    {
      "ref": "zai-org/GLM-4.5-Air-FP8",
      "family": "glm-4-5",
      "member": "air",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-4.5-Air-FP8",
      "hub_repo": "zai-org/GLM-4.5-Air-FP8",
      "publisher": "zai-org",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 107348,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.5-Air-FP8",
        "fetched_at": "2026-11-10T13:25:20.254Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.5-air",
        "ids": [
          "z-ai/glm-4.5-air"
        ],
        "usd_per_m_in": 0.2,
        "usd_per_m_out": 1.1,
        "usd_per_m_cached_in": 0.03,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 42.3166,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "zai-org/GLM-4.5-FP8",
      "family": "glm-4-5",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-4.5-FP8",
      "hub_repo": "zai-org/GLM-4.5-FP8",
      "publisher": "zai-org",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 344551,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.5-FP8",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.5",
        "ids": [
          "z-ai/glm-4.5"
        ],
        "usd_per_m_in": 0.6,
        "usd_per_m_out": 2.2,
        "usd_per_m_cached_in": 0.11,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 268.4412,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "zai-org/GLM-4.6-FP8",
      "family": "glm-4-6",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-4.6-FP8",
      "hub_repo": "zai-org/GLM-4.6-FP8",
      "publisher": "zai-org",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 344551,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.6-FP8",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.6",
        "ids": [
          "z-ai/glm-4.6"
        ],
        "usd_per_m_in": 0.5,
        "usd_per_m_out": 2,
        "usd_per_m_cached_in": 0.1,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 268.4412,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "zai-org/GLM-4.6V-FP8",
      "family": "glm-4-6",
      "member": "v",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-4.6V-FP8",
      "hub_repo": "zai-org/GLM-4.6V-FP8",
      "publisher": "zai-org",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 104908,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.6V-FP8",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.6v",
        "ids": [
          "z-ai/glm-4.6v"
        ],
        "usd_per_m_in": 0.3,
        "usd_per_m_out": 0.9,
        "usd_per_m_cached_in": 0.055,
        "context_length": 131072,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 4.81,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 40.936,
          "tariff_basis": "bootstrap",
          "verdict": "estimated"
        }
      }
    },
    {
      "ref": "zai-org/GLM-4.7-FP8",
      "family": "glm-4-7",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-4.7-FP8",
      "hub_repo": "zai-org/GLM-4.7-FP8",
      "publisher": "zai-org",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 345290,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-4.7-FP8",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-4.7",
        "ids": [
          "z-ai/glm-4.7"
        ],
        "usd_per_m_in": 0.4,
        "usd_per_m_out": 1.75,
        "usd_per_m_cached_in": 0.08,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.46,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 268.8531,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "zai-org/GLM-5-FP8",
      "family": "glm-5",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-5-FP8",
      "hub_repo": "zai-org/GLM-5-FP8",
      "publisher": "zai-org",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 721147,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5-FP8",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5",
        "ids": [
          "z-ai/glm-5"
        ],
        "usd_per_m_in": 1,
        "usd_per_m_out": 3.2,
        "usd_per_m_cached_in": 0.2,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 997.7124,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "zai-org/GLM-5.1-FP8",
      "family": "glm-5-1",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-5.1-FP8",
      "hub_repo": "zai-org/GLM-5.1-FP8",
      "publisher": "zai-org",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 721147,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.1-FP8",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.1",
        "ids": [
          "z-ai/glm-5.1"
        ],
        "usd_per_m_in": 1.4,
        "usd_per_m_out": 4.4,
        "usd_per_m_cached_in": 0.26,
        "context_length": 204800,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 997.7124,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "zai-org/GLM-5.2-FP8",
      "family": "glm-5-2",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-5.2-FP8",
      "hub_repo": "zai-org/GLM-5.2-FP8",
      "publisher": "zai-org",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 720626,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.2-FP8",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.2",
        "ids": [
          "z-ai/glm-5.2"
        ],
        "usd_per_m_in": 0.007,
        "usd_per_m_out": 6.8,
        "usd_per_m_cached_in": 0.006,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": null,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 997.1067,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "zai-org/GLM-5.3",
      "family": "glm-5-3",
      "member": "base",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-5.3",
      "hub_repo": "zai-org/GLM-5.3",
      "publisher": "zai-org",
      "derivative": false,
      "cohort": 2,
      "rung": 0,
      "disk_mib": 720626,
      "active_share": 1,
      "licence": {
        "id": "other",
        "name": "glm-5.3",
        "class": "restricted",
        "serve_scopes": [
          "own",
          "gramx"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.3",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.3",
        "ids": [
          "z-ai/glm-5.3",
          "z-ai/glm-5.3:batch"
        ],
        "usd_per_m_in": 0.084,
        "usd_per_m_out": 3.6,
        "usd_per_m_cached_in": 0.079,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 2,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 0.7,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 997.1067,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    },
    {
      "ref": "zai-org/GLM-5.3-Flash",
      "family": "glm-5-3",
      "member": "flash",
      "quant": "FP8",
      "engine": "vllm",
      "modality": "text",
      "base": "zai-org/GLM-5.3-Flash",
      "hub_repo": "zai-org/GLM-5.3-Flash",
      "publisher": "zai-org",
      "derivative": false,
      "cohort": 0,
      "rung": 0,
      "disk_mib": 313126,
      "active_share": 1,
      "licence": {
        "id": "mit",
        "name": null,
        "class": "open",
        "serve_scopes": [
          "own",
          "gramx",
          "open-mesh"
        ],
        "gated": false,
        "source": "https://huggingface.co/api/models/zai-org/GLM-5.3-Flash",
        "fetched_at": "2026-11-10T13:40:45.998Z"
      },
      "seasons": [],
      "api": {
        "openrouter_id": "z-ai/glm-5.3-flash",
        "ids": [
          "z-ai/glm-5.3-flash",
          "z-ai/glm-5.3-flash:batch"
        ],
        "usd_per_m_in": 0.0399,
        "usd_per_m_out": 0.0784,
        "usd_per_m_cached_in": 0.03,
        "context_length": 1048576,
        "free_variant": false,
        "batch_usd_per_m_out": 0.18,
        "source": "https://openrouter.ai/api/v1/models",
        "fetched_at": "2026-11-15T19:26:02.086Z"
      },
      "no_hosted_api": false,
      "best": null,
      "classes": {
        "linux/amd64/nvidia/64g": {
          "proved": false,
          "grams": 0,
          "seasons": [],
          "tokens_per_sec_best": null,
          "effective_ctx_best": null,
          "joules_per_token_best": null,
          "first_at": null,
          "last_at": null,
          "caps": [],
          "est_tokens_per_sec": 1.61,
          "est_basis": "machines of this class",
          "est_usd_per_m_out": 227.5468,
          "tariff_basis": "bootstrap",
          "verdict": "too slow here (estimate)"
        }
      }
    }
  ]
}
